NATO's Air War For Kosovo - A Strategic and Operational Assessment-Rand Publishing (2001)

RAND MR1365-1.
Nis
l
Raska l l
Blace
l
Sjenica K
L
O
P
Prokuplje
A
O
l
l
Zitorada
N
IK Doljevac
l
l Brestovac l
S
l
Kursumlija
Novi
A
Pazar
N
S e r b i a Bojnik
D
l
l
Leskovac
Zˇ
l
Tutin
A
Lebane l
Montenegro l Kosovska
Mitrovica Podujevo
l
K
Zubin Potok Medvedja l

l
Berane ˇ
Rozaj
l
l
ˇ
Vucitrn
l
Istok l K
O
Srbica l S
l
l
Poljance O G
l Djurakovac V O
Pec Vitomirica O L J
l Pristina A K
Klina Glogovac Kosovo l Novo
l
l Poljec Brdo Kosovska
M E
l
l Gracanica l Kamenica
l
Plav l
P
l
Lapusnik
l
Janjevo Vranje l
T
Decani l Lipljan
O
o
L
N
.A
J
H I
LB
Malisevo l
E
l
AN
Junik Gnjilane
l l
I
AN
Bujanovac
J A
AL
PS
l
Bajram Orahovac
l
Curril Djakovica l
Suva Reka Urosevac

l
l Presevo
Vitina l
Kosovo
Kacanik
l
A L B A N I A
CR
l Prizren
NA
Shalqin l
l
Blace Kumanovo l
GO
Kukes
RA
l
Dragas A
IN
l
N
A Tetovo l
L
P
l
Brod Skopje
R
l
Kamenjane
SA
Gostivar l
MACEDONIA
l
Titov
Veles
Map of Kosovo
PREFACE
On March 24, 1999, NATO embarked on a 78-day air war aimed at

compelling the government of Yugoslavia and its elected president,
Slobodan Milosevic, to halt and reverse the human rights abuses that
were being committed by armed Serbs against the ethnic Albanian
majority living in Yugoslavia’s Serbian province of Kosovo. That ef-
fort, called Operation Allied Force, ended on June 9 after Milosevic
finally acceded to NATO’s demands and a withdrawal of Serb forces
from Kosovo had begun. The air war was a first of that magnitude for
NATO and represented the third largest strategic application of air
power by the United States since World War II, exceeded only by the
Vietnam War and Operation Desert Storm in scale and intensity.
With a view toward capturing the many useful insights to be ex-

tracted from that experience, the U.S. Air Force chief of staff, General
Michael Ryan, asked Headquarters United States Air Forces in
Europe (Hq USAFE) shortly after Allied Force ended to establish
a studies and analysis office (USAFE/SA) to manage all USAF-
sponsored assessments of the air war. The director of that office,
Brigadier General John Corley, in turn asked RAND’s Project AIR
FORCE to contribute to the assessment effort across a wide spectrum
of topics, ranging from individual platform and systems performance
to command and control, operational support, strategy and plan-
ning, and other considerations bearing on the air war’s effectiveness.
This book examines the conduct and results of Operation Allied

Force at the strategic and operational levels. An earlier and less de-
veloped version appeared as a chapter in the author’s previous book
The Transformation of American Air Power, which was published by
Cornell University Press in September 2000. The research docu-
v
vi NATO’s Air War for Kosovo: A Strategic and Operational Assessment
mented herein was carried out in Project AIR FORCE’s Strategy and
Doctrine Program and was completed in August 2001. All
photographs included in this study were provided by the U.S.
Department of Defense. The book should be of interest to USAF
officers and other members of the U.S. national security community
concerned with strategy and force employment issues raised by
NATO’s air war for Kosovo and with the implications of that
experience for force development, air power doctrine, and concepts
of operations for joint and coalition warfare.
Other documents published in this series currently include the fol-

lowing:
MR-1279-AF, Command and Control and Battle Management: Expe-

riences from the Air War over Serbia, James E. Schneider, Myron
Hura, Gary McLeod (Government publication; not releasable to the
general public)
MR-1326-AF, Aircraft Weapon Employment in Operation Allied Force,

William Stanley, Carl Rhodes, Robert Uy, Sherrill Lingel (Government
publication; not releasable to the general public)
MR-1351-AF, The Conflict over Kosovo: Why Milosevic Decided to

Settle When He Did, Stephen Hosmer
MR-1391-AF, European Contributions to Operation Allied Force:

Implications for Transatlantic Cooperation, John E. Peters, Stuart
Johnson, Nora Bensahel, Timothy Liston, Traci Williams
DB-332-AF, Aircraft Survivability in Operation Allied Force, William
Stanley, Sherrill Lingel, Carl Rhodes, Jody Jacobs, Robert Uy
(Government publication; not releasable to the general public)
Topics examined in series documents nearing completion include:
• Supporting Expeditionary Aerospace Forces: Lessons from the

Air War Over Serbia
• Lessons Learned from Operation Allied Force Tanker Operations
Preface vii
Project AIR FORCE

Project AIR FORCE, a division of RAND, is the Air Force’s federally
funded research and development center (FFRDC) for studies and
analysis. It provides the USAF with independent analyses of policy
alternatives affecting the deployment, employment, combat readi-
ness, and support of current and future air and space forces. Re-
search is performed in four programs: Aerospace Force Develop-
ment; Manpower, Readiness, and Training; Resource Management;
and Strategy and Doctrine.
FIGURES
2.1. Allied Force Area of Operations .................................... 13

3.1. U.S. and Allied Aircraft Contributions .......................... 33
3.2. In-Theater Aircraft Buildup .......................................... 35
3.3. U.S. and Allied Sorties Flown ....................................... 50
3.4. USAF Sortie Breakdown by Aircraft Type ...................... 63
3.5. U.S. and Allied Ground-Attack Munitions
Expended .................................................................... 64
3.6. U.S. and Allied Munitions Expenditures by Type ........... 65
3.7. Total Numbers of Munitions Expended ........................ 66
5.1. U.S. Precision and Nonprecision Munitions
Expended .................................................................... 88
6.1. Enemy SAM Launches Reported .................................. 109
6.2. HARM Expenditures by Target Type ............................. 110
6.3. Short Tonnage Delivered by USAF Airlift ...................... 159
6.4. USAF Aircraft Types Employed .................................... 174
7.1. Operation Allied Force Planning and
Implementation .......................................................... 189
7.2. U.S. and Allied Organization for Allied Force ................ 208
8.1. Refugee Flow .............................................................. 227
xi
SUMMARY
Between March 24 and June 9, 1999, NATO, led by the United States,
conducted an air war against Yugoslavia in an effort to halt and re-
verse the human-rights abuses that were being committed against
the citizens of its Kosovo province by Yugoslavia’s president, Slobo-
dan Milosevic. That 78-day air war, called Operation Allied Force,
represented the third time during the 1990s in which air power
proved pivotal in determining the outcome of a regional conflict. Yet
notwithstanding its ultimate success, what began as a hopeful
gambit for producing Milosevic’s quick compliance soon devolved,
for a time at least, into a seemingly ineffectual bombing experiment
with no clear end in sight. Not only was the operation’s execution
hampered by uncooperative weather and a surprisingly resilient op-
ponent, it was further afflicted by persistent hesitancy on the part of
U.S. and NATO political leaders and sharp differences of opinion
within the most senior U.S. military command element over the most
effective way of applying allied air power against Serb assets. More-
over, the plan ultimately adopted ruled out any backstopping by al-
lied ground troops because of concerns over the potential for a land
invasion to generate unacceptable casualties and the consequent low
likelihood of mustering the needed congressional and allied support
for such an option. All planning further assumed that NATO’s most
crucial vulnerable area was its continued cohesion. Therefore, any
target or attack tactic deemed even remotely likely to undermine that
cohesion, such as the loss of friendly aircrews, excessive collateral
damage, or anything else that might weaken domestic support, was
to be most carefully considered, if not avoided altogether. All of that,
however unavoidable some aspects of it may have been, made
xiii
xiv NATO’s Air War for Kosovo: A Strategic and Operational Assessment
NATO’s air war for Kosovo a step backward in efficiency when com-
pared to the Desert Storm air campaign.
WHY MILOSEVIC GAVE UP WHEN HE DID

We may never know for sure what mix of pressures and inducements
ultimately led Milosevic to admit defeat. Yet why he gave in and why
he did so when he did are by far the most important questions about
the operation’s experience, since the answers, insofar as they are
knowable, may help illuminate the coercive dynamic that ultimately
swung the air war’s outcome.
One can, of course, insist that air power alone was the cause of Milo-
sevic’s capitulation in the tautological sense that Allied Force was an
air-only operation and that in its absence, there would have been no
reason for believing that he would have acceded to NATO’s de-
mands. Yet as crucial as the 78-day bombing effort was in bringing
Milosevic to heel, one should be wary of any intimation that NATO’s
use of air power produced a successful result for the alliance without
any significant contribution by other factors. For example, beyond
the obvious damage that was being caused by NATO’s air attacks and
the equally obvious fact that NATO could have continued bombing
both indefinitely and with virtual impunity, another likely factor be-
hind Milosevic’s capitulation was the fact that the sheer depravity of
Serbian conduct in Kosovo had stripped the Yugoslavian leader of
any remaining vestige of international support, including, in the end,
from his principal backers in Moscow.
On top of that was the sense of walls closing in that Milosevic must
have had when he was indicted as a war criminal by a UN tribunal
only a week before his loss of Moscow’s support. Yet a third factor
may have been the mounting pressure from Milosevic’s cronies
among the Yugoslav civilian oligarchy, prompted by the continued
bombing of military-related industries, utilities, and other infrastruc-
ture targets in and around Belgrade in which they had an economic
stake and whose destruction increasingly threatened to bankrupt
them.
Finally, one must take into account what Milosevic no doubt per-
ceived, rightly or wrongly, to have been the possibility of an eventual
Summary xv
NATO ground invasion. Whatever NATO’s declared stance on the

ground-war issue may have been, its actions as the bombing pro-
gressed spoke louder than its words. By the end of May, it had be-
come clear that the alliance was beginning to come to grips with the
necessity for a ground intervention of some sort if the bombing did
not produce the desired result soon. Milosevic knew that and fully
appreciated what it meant for his political fortunes.
Some, however, have made more of that fact than the evidence war-
rants. In the early wake of the successful conclusion of Allied Force,
revisionist claims began emanating from some quarters suggesting
that the air effort had been totally ineffective and that, in the end, it
had been Milosevic’s fear of a NATO ground invasion that had in-
duced him to capitulate. Those claims defy believability because any
NATO ground invasion, however probable it may have been in the
end, would have taken months, at a minimum, to prepare for and
successfully mount.
In contrast, Milosevic was living with the daily reality of an increas-

ingly brutal air war that was showing no sign of abating. Although
the effort to find and attack dispersed and hidden enemy forces in
Kosovo was consuming the preponderance of ground-attack sorties
while accomplishing little by way of tangible results on the ground,
more and more infrastructure targets were also being approved and
struck every day. Accordingly, there is no basis for concluding that
the mere possibility of an eventual land invasion somehow over-
shadowed the continuing reality of NATO’s air attacks as the preemi-
nent consideration accounting for Milosevic’s decision to capitulate.
The bombing ultimately persuaded Milosevic that NATO not only
would not relent, but also was determined to prevail and had both
the technical and political wherewithal to do so. By the same token,
given the incapacity of Serb air defenses to shoot down significant
numbers of allied aircraft, the bombing further convinced him that
his own defeat was inevitable sooner or later.
FRICTION AND OPERATIONAL PROBLEMS

Although NATO’s use of air power in Allied Force must, in the end, be
adjudged a success, some troubling questions arose well before the
operation’s favorable outcome over a number of disconcerting
xvi NATO’s Air War for Kosovo: A Strategic and Operational Assessment
problems that were encountered along the way. Among those

arousing the greatest concern were the following:
• Assessed deficiencies in the suppression of enemy air defenses

(SEAD).
• Locating, identifying, and engaging dispersed and hidden enemy
light infantry forces in Kosovo.
• Inadvertent civilian casualties.
In contrast to the far more satisfying SEAD experience in Desert

Storm, the effort to neutralize Serb air defenses did not go nearly as
well as hoped. The Serbs kept most of their surface-to-air missiles
(SAMs) in standby mode with their radars not emitting, prompting
concern that they were attempting to draw NATO aircraft down to
lower altitudes where they could be more easily engaged. The un-
derstandable reluctance of enemy SAM operators to emit and thus
render themselves cooperative targets made them much harder to
find and attack, forcing allied aircrews to remain constantly alert to
the radar-guided SAM threat. By the same token, the enemy’s heavy
man-portable air defense system (MANPADS) and antiaircraft
artillery (AAA) threat forced allied aircrews to bomb from above
15,000 ft, for the most part, to remain outside their lethal envelopes.
Moreover, because of the mountainous terrain of Kosovo, the
moving target indicator and synthetic aperture radar aboard the E-8
Joint STARS could not detect targets at oblique look angles, although
the sensors carried by the higher-flying U-2 often compensated for
this shortfall. On the plus side, although enemy SAM operators
aggressively attempted to engage allied aircraft throughout the air
war, superior allied SEAD operations forced them to employ
emission control and mobility tactics to enhance their survivability,
which significantly decreased their effectiveness. In the end, only
two NATO aircraft were brought down by enemy fire, thanks to allied
reliance on electronic jamming, the use of towed decoys, and
countertactics to negate enemy surface-to-air defenses. However,
NATO never fully succeeded in neutralizing the enemy’s radar-
guided SAM threat, even though no areas of enemy territory were
denied.
Still another disappointment centered on what turned out to be

NATO’s almost completely ineffective efforts to engage mobile en-
Summary xvii
emy troops operating in Kosovo. That disappointment underscored

the limits of conducting air strikes against dispersed enemy forces
hiding in favorable terrain in the absence of a supporting allied
ground threat. Had Serb commanders any reason to fear a NATO
ground invasion, they would have had little alternative but to posi-
tion their tanks to cut off roads and other avenues of attack, thus
making their forces more easily targetable by NATO air power. In-
stead, having dispersed and hidden their tanks and armored person-
nel carriers, Serb army and paramilitary units were free to go in with
just a few troops in a single vehicle to terrorize a village in connection
with their ethnic cleansing campaign.
Senior civilian defense officials and U.S. Air Force leaders freely con-
ceded after the Serbian withdrawal that the problems encountered
by the largely failed effort against fielded enemy forces reflected real
challenges for the effective application of air power posed by such
impediments as trees, mountains, poor weather, and an enemy
ground force that is permitted the luxury of dispersing and hiding
rather than concentrating to maneuver to accomplish its mission.
Yet while it was essential for NATO to try its best to keep Serb forces
pinned down and incapable of operating at will, the majority of the
sorties devoted to finding and attacking enemy troops in Kosovo en-
tailed an inefficient and ineffective use of munitions and other valu-
able assets. That said, the targeting of enemy ground forces operat-
ing within Kosovo was an inescapable political necessity, considering
that those forces were responsible for committing the ethnic cleans-
ing acts that NATO had vowed to stop. Failure to target those forces
would almost certainly have caused the bombing effort to lose cred-
ibility in the eyes of the NATO civilian leadership.
Pressures to avoid civilian casualties and unintended damage to

nonmilitary structures were greater in Allied Force than in any previ-
ous combat operation involving U.S. forces. Nevertheless, there were
recurrent instances throughout the air war of unintended damage
caused either by errant NATO munitions or by mistakes in targeting,
including a dozen highly publicized incidents in which civilians were
accidentally killed. One such bombing error resulted in part from
constraints imposed by the requirement that NATO aircrews remain
above 15,000 ft to avoid the most lethal enemy threats, making visual
discrimination between military and civilian traffic more than rou-
tinely difficult. Another contributing factor was the occasional ten-
xviii NATO’s Air War for Kosovo: A Strategic and Operational Assessment
dency of allied aircrews to maneuver their aircraft in such a way as to

put clouds within the targeting pod’s field of view between the air-
craft and the target, thus blocking the laser beam illuminating the
target and depriving the weapon of guidance. Moreover, Serb forces
often used civilians as human shields in an effort to deter NATO from
attacking military vehicles. The extraordinary media attention given
to these events bore ample witness to what can happen when zero
noncombatant casualties becomes not only a goal of strategy but
also the international expectation. Thanks to unrealistic efforts to
treat the normal friction of war as avoidable human error, every oc-
currence of unintended collateral damage became overinflated as
front-page news and treated as a blemish on air power’s presumed
ability to be consistently precise.
LAPSES IN STRATEGY AND IMPLEMENTATION

NATO’s leaders also had little to congratulate themselves about
when it came to the way in which the air war was planned and car-
ried out. There was a dominant sense among participants and ob-
servers alike that the desultory onset of Allied Force and its later
slowness to register effects reflected some fundamental failures of
allied leadership and strategy choice. In contrast to the relatively
seamless performance by the coalition in Desert Storm, what un-
folded during NATO’s air war for Kosovo was a highly dissatisfying
application of air power, which showed not only the predictable fits
and starts of trying to prosecute an air operation through an alliance
of 19 members bound by a unanimity rule, but also some failures
even within the operation’s U.S. component to make the most of
what air power had to offer within the limits of the effort’s political
constraints.
To begin with, the conduct of the air war as an allied effort came at
the cost of a flawed strategy that was further hobbled by the manifold
inefficiencies that were part and parcel of conducting combat opera-
tions by consensus. In addition to the natural friction created by
NATO’s approach to target approval, the initial reluctance of its
political leaders to countenance a more aggressive air campaign in
terms of target numbers and force size failed completely to capitalize
on air power’s potential for taking down entire systems of enemy
capability simultaneously. Further compounding the inefficiency of
Summary xix
this multistage and circuitous process, two parallel but separate

mechanisms for mission planning and air tasking were used. Any
U.S.-specific systems involving special sensitivities, such as the B-2,
F-117, and cruise missiles, were allocated by U.S. European Com-
mand (USEUCOM) rather than by NATO, and the Combined Air
Operations Center (CAOC) maintained separate targeting teams for
USEUCOM and NATO strike planning.
Because NATO had initially hoped that the operation would last only
a few days, it failed to establish a smoothly running mechanism for
target development and review until late April. Once NATO’s going-
in assumption proved hollow, a frenetic rush ensued to come up
with additional target nominations that could be more quickly and
easily approved by NATO’s political authorities. Even then, there
was little by way of a consistently applied strategy behind the target
development process. Most of the attack planning done throughout
the air war was not driven by desired effects, but rather entailed
simply parceling out sortie and munitions allocations by target cate-
gory as individual targets were approved, without much considera-
tion given to how a target’s neutralization might contribute toward
advancing the overall objectives of the air war.
It was not only the alliance-induced friction that helped make for an
inefficient bombing effort. As Allied Force unfolded, it became in-
creasingly clear that even the U.S. military component was divided in
a high-level struggle over the most appropriate targeting strategy—
reminiscent of the feuding that had occurred nine years earlier be-
tween the Army’s corps commanders and the joint force air compo-
nent commander (JFACC), then–Lieutenant General Charles Horner,
over the ownership and control of air operations in Desert Storm.
Once the initial hope that Milosevic would fold within a few days af-
ter the bombing started proved groundless, NATO was forced into a
scramble to develop an alternative strategy. The immediate result
was an internecine battle between the Supreme Allied Commander
in Europe, U.S. Army General Wesley Clark, and his air component
commander, USAF Lieutenant General Michael Short, over where
the air attacks should be primarily directed. Short maintained that
the most effective use of allied air power would be to pay little heed
to dispersed Serbian forces in Kosovo and to concentrate instead on
infrastructure targets in and around Belgrade, including key electri-
cal power plants and government ministries. However, Clark in-
xx NATO’s Air War for Kosovo: A Strategic and Operational Assessment
sisted, as was his command prerogative, upon concentrating on elu-

sive enemy ground troops in Kosovo, and this targeting emphasis
prevailed throughout most of the air war.
Despite the success of Allied Force in the end, one misjudgment of

near-blunder proportions came close to saddling the United States
and NATO with a costly and embarrassing failure. The worst call by
NATO’s leaders was their assumption that what had worked for
Bosnia would work for Kosovo and their resultant failure to appreci-
ate the special importance of Kosovo to the Serbs and its criticality to
Milosevic’s survival in power. Fortunately for the allies, their faulty
assessment was not a show-stopper—although it could have been if
Milosevic had refrained from launching his ethnic cleansing cam-
paign and instead merely hunkered down to wait out the bombing in
a win-or-lose contest of wills with NATO. Had he done so, he could
have threatened the long-term viability of the alliance. Fortunately
for the success of Allied Force, by opting instead to accelerate his
ethnic cleansing of Kosovo, he helped unite Western opinion behind
NATO and left NATO with no choice but to dig in for the long haul,
not only to secure an outcome that would allow for the repatriation
of nearly a million displaced Kosovars, but also to ensure its contin-
ued credibility as an alliance.
NATO’S AIR WAR IN PERSPECTIVE

Operation Allied Force was the most intense and sustained military
operation to have been conducted in Europe since the end of World
War II. It represented the first extended use of military force by
NATO, as well as the first major combat operation conducted for
humanitarian objectives against a state committing atrocities within
its own borders. It was the longest U.S. combat operation to have
taken place since the war in Vietnam. At a price tag of more than
$3 billion, it was also a notably expensive one. Yet in part because of
that investment, it turned out to have been an unprecedented exer-
cise in the discriminate use of force on a large scale. In all, out of
some 28,000 high-explosive munitions expended over the opera-
tion’s 78-day course, no more than 500 noncombatants died as a di-
rect result of errant attacks.
After the bombing ended, the predominant tendency among most

outside observers was to characterize that effort as a watershed
Summary xxi
achievement for air power. Yet with all due respect for the un-
matched professionalism of the aircrews who actually carried out the
air war, it is hard to accept that characterization as the proper con-
clusion to be drawn from Allied Force. To be sure, there is much to
be said of a positive nature about NATO’s air war for Kosovo. To be-
gin with, it did indeed represent the first time in which air power co-
erced an enemy leader to yield with no friendly land combat action
whatsoever. This does not mean that air power can now “win wars
alone” or that the air-only strategy ultimately adopted by NATO’s
leaders was the wisest choice available to them. Yet the fact that air
power prevailed on its own despite the multiple drawbacks of a re-
luctant administration, a divided Congress, an indifferent public, a
potentially fractious alliance, a determined opponent, and—not
least—the absence of a credible NATO strategy surely testified that
the air instrument has come a long way in recent years in its relative
combat leverage compared to that of other force elements in joint
warfare.
The two most important accomplishments of the air war occurred at

the strategic level and had to do with the performance of the alliance
as a combat collective. First, despite the suggestion of some critics to
the contrary, NATO clearly prevailed over Milosevic. Although
NATO’s air strikes were unable to halt Milosevic’s ethnic cleansing
campaign before it had been essentially completed, they did succeed
in completely reversing its effects in the early aftermath of the cease-
fire by forcing Milosevic to accept NATO’s demands. Second, NATO
showed that it could operate successfully under pressure as an al-
liance, even in the face of constant hesitancy and reluctance on the
part of many of the member states’ political leaders. For all the air
effort’s fits and starts and the manifold frustrations they caused, the
alliance earned justified credit for having done remarkably well in a
uniquely challenging situation. In seeing Allied Force to a successful
conclusion, NATO did something that it had been neither created
nor configured to do.
Despite these accomplishments, there were enough disappoint-

ments to suggest that instead of basking in the glow of air power’s
successful performance, air warfare professionals should give careful
thought to the hard work that still needs to be done if air power’s
fullest potential in joint warfare is to be realized. As with the opera-
tion’s successes, the biggest failings of Allied Force occurred in the
xxii NATO’s Air War for Kosovo: A Strategic and Operational Assessment
realm of strategy and execution. First, despite its successful outcome

and through no fault of allied airmen, the bombing effort was a sub-
optimal application of air power. The incremental plan chosen by
NATO’s leaders risked squandering much of the capital that had
been built up in air power’s account ever since its success in Desert
Storm nearly a decade before. Almost without question, the first
month of underachievement in the air war convinced Milosevic that
he could ride out the NATO attacks. Indeed, the way Operation Al-
lied Force commenced violated two of the most enduring axioms of
military practice: the importance of achieving surprise and the criti-
cality of keeping the enemy unclear as to one’s intentions. The ac-
ceptance by NATO’s leaders of a strategy that ruled out a ground
threat from the very start and envisaged only gradually escalating air
strikes to inflict pain was a guaranteed recipe for downstream trou-
ble, even though it was the only strategy that, at the time, seemed
politically workable.
Although the manner in which the air war was conducted fell short of
the ideal use of air power, it suggested that gradualism may be here
to stay if U.S. leaders ever again intend to fight with coalition part-
ners for marginal or amorphous interests. Insofar as gradualism
promises to be the wave of the future, it suggests that airmen will
need to discipline their natural inclination to bridle whenever politi-
cians moderate the application of a doctrinally pure campaign strat-
egy and to recognize and accept instead that political considerations,
after all, determine—or should determine—the way in which cam-
paigns and wars are fought. This does not mean that military leaders
should surrender to political pressures without first making their
best case for using force in the most effective and cost-minimizing
way. It does, however, stand as an important reminder that war is ul-
timately about politics and that civilian control of the military is an
inherent part of the democratic tradition. Although air warfare pro-
fessionals, like all other warfighters, are duty-bound to try to per-
suade their civilian superiors of the merits of their recommenda-
tions, they also have a duty to live with the hands they are dealt and
to bend every effort to make the most of them in an imperfect world.
It follows that civilian leaders at the highest levels have an equal obli-
gation to try to stack the deck in such a way that the military has the
best possible hand to play and the fullest possible freedom to play it
to the best of its ability. This means expending the energy and politi-
Summary xxiii
cal capital needed to develop and enforce a strategy that maximizes

the probability of military success. That was not done by the vast
majority of the topmost civilian leaders on either side of the Atlantic
in Allied Force.
On the plus side, the operation’s successful outcome—despite its

many frustrations—suggests that U.S. air power may now have be-
come capable enough to underwrite a strategy of incremental esca-
lation irrespective of that strategy’s inherent inefficiencies. What
made the gradualism of Allied Force more bearable than that of the
earlier war in Vietnam was that NATO’s advantages in stealth, preci-
sion standoff attack, and electronic warfare meant that it could fight
a one-sided war against Milosevic with near-impunity and achieve its
desired result, even if not in the most ideal way. That was not an op-
tion when U.S. air power was a less developed tool than it is today.
One of the most important realizations to emerge from Allied Force

had to do with the opportunity costs incurred by NATO’s anemic
start of its air attacks without an accompanying ground threat. They
included the following:
• A failure to exploit air power’s inherent shock potential and to

instill in Milosevic an early fear of worse consequences yet to
come.
• The encouragement the initial lack of a NATO ground threat gave
enemy troops to disperse and hide while they had time.
• The virtual carte blanche that lack gave Milosevic for accelerated
atrocities in Kosovo.
• The relinquishment of the power of the initiative to the enemy.
These problems identified by the Allied Force experience suggest an

important corrective to the unending argument between airmen and
land combatants over the relative combat merits of air power versus
“boots on the ground.” Although Allied Force reconfirmed that
friendly ground forces no longer need to be inexorably committed to
combat early, it also reconfirmed that air power in many cases can-
not perform to its fullest potential without the presence of a credible
ground component in the overall allied strategy.
xxiv NATO’s Air War for Kosovo: A Strategic and Operational Assessment
ON THE USES AND ABUSES OF AIR POWER

Viewed in hindsight, the most remarkable thing about Operation Al-
lied Force is not that it defeated Milosevic in the end, but rather that
air power prevailed despite a NATO leadership that was unwilling to
take major risks and an alliance that held together only with often
paralyzing drag. Lesson One from both Vietnam and Desert Storm
should have been that one must not commit air power in “penny
packets,” as the British say, to play less-than-determined games with
the risk calculus of the other side. Although it can be surgically pre-
cise when precision is called for, air power is, at bottom, a blunt in-
strument designed to break things and kill people in pursuit of clear
and militarily achievable objectives.
To admit that gradualism may be the wave of the future for any U.S.
involvement in coalition warfare is hardly to accept that it is any
more justifiable from a military point of view for that reason alone.
Quite to the contrary, the incrementalism of NATO’s air war for
Kosovo, right up to its very end, involved a potential price that went
far beyond the loss of valuable aircraft and other expendables for
questionable gain. It risked frittering away the hard-earned reputa-
tion for effectiveness that U.S. air power had finally earned for itself
in Desert Storm after more than three years of unqualified misuse
over North Vietnam a generation earlier.
As the Gulf War experience showed, and as both Deliberate Force

and Allied Force ultimately reaffirmed, U.S. air power as it has
evolved since the mid-1970s can do remarkable things when em-
ployed with determination in support of a campaign whose intent is
not in doubt. Yet to conjure up the specter of “air strikes”—NATO or
otherwise—in an effort to project an appearance of “doing some-
thing” without a prior weighing of intended effects or likely conse-
quences is to run the risk of getting bogged down in an operation
with no plausible theory of success. After years of false promises by
its most outspoken prophets, air power has become an unprecedent-
edly capable instrument of force employment in joint warfare. Even
in the best of circumstances, however, it can never be more effective
than the strategy it is intended to support.
ACKNOWLEDGMENTS
This study has benefited from insights gained from conversations

with numerous participants in Operation Allied Force, including
Admiral James Ellis, Jr., USN, commander in chief, Allied Forces
Southern Europe and Joint Task Force commander, Operation Noble
Anvil; General John Jumper, USAF, former commander, United
States Air Forces in Europe and now U.S. Air Force chief of staff;
Lieutenant General Michael Short, USAF (Ret.), former commander,
Allied Air Forces Southern Europe and Joint Force Air Component
Commander, Operational Noble Anvil; Vice Admiral Daniel Murphy,
Jr., USN, commander, 6th Fleet; Captain Dave Maxwell, USN, direc-
tor of operations, and Captain Tony Cothron, USN, director of intel-
ligence, 6th Fleet; Lieutenant General Charles Wald, USAF, former
deputy director, J-5, Joint Staff and now commander, 9th Air Force;
Major General (select) Dan Leaf, former commander, 31st Air
Expeditionary Wing at Aviano Air Base, Italy, and now director of
operational requirements, Hq USAF; Air Chief Marshal Sir John Day,
RAF, the UK Ministry of Defense’s director of operations during
Allied Force and now commander in chief, RAF Strike Command; Air
Vice Marshal Andrew Vallance, RAF, former chief of staff, NATO
Reaction Force (Air); Major General P. J. M. Godderij, deputy com-
mander in chief, Royal Netherlands Air Force; Captain C. J. Heatley,
USN (Ret.), former commander, Joint Warfare Analysis Center;
Colonel Jeff Eberhart, commander, and Lieutenant Colonel Ray
Dissinger, deputy commander, 31st Operations Group, Aviano Air
Base, Italy; and Lieutenant Colonel Steve Schraeder, commander,
510th Fighter Squadron at Aviano.
xxv
xxvi NATO’s Air War for Kosovo: A Strategic and Operational Assessment
I am also indebted to Major General Eitan Ben-Eliahu, then–

commanding officer of the Israeli Air Force (IAF), who invited me to
brief the highlights of an earlier version of this book and to receive
feedback from the entire senior IAF leadership at a special
roundtable session convened during his weekly staff meeting at IAF
headquarters in Tel Aviv in October 1999. In addition, I acknowledge
Thomas Henriksen, associate director of the Hoover Institution at
Stanford University, who organized a seminar for me at Hoover in
April 2000, attended by Hoover’s military fellows and former Secre-
tary of Defense William Perry, to critique an earlier draft of this study;
Major General Gary Dylewski, then-commander, and Colonel Bob
Bivins, then–director of operations, Air Force Space Warfare Center,
for having briefed me on the contribution of space systems to
Operation Allied Force; General Patrick Gamble, commander, and
Lieutenant General (select) Thomas Waskow, director of air and
space operations, Pacific Air Forces, for sharing their insights into the
Allied Force experience; Major Marshall Denney, USMC, Tactics Di-
vision, Marine Aviation Weapons and Tactics Squadron 1, for de-
scribing some of the highlights of EA-6B operations; Major General
Leroy Barnidge, former commander, 509th Bomb Wing and later vice
commander, 9th Air Force, for his observations on B-2 employment
during Allied Force; and Loren Timm of Lockheed Martin Corpora-
tion for providing me with useful material on F-16 operations.
Most important in this respect, I thank General Gregory Martin,

commander, United States Air Forces in Europe, for having put the
extensive USAFE Air War Over Serbia (AWOS) data collection at my
complete disposal, and Lieutenant Colonel Richard Sargent, director,
operations studies and analysis, Hq USAFE, and Edward Ballanco,
Hq USAFE, for carefully vetting the penultimate draft of this study
and helping me to make the most of the unclassified documentation
that General Martin made available to me. I also thank Captain John
Wilbourne, 492nd Fighter Squadron, RAF Lakenheath, England, for
his special effort to put me in touch with both individuals and re-
sources at Lakenheath with a direct connection to 48th Fighter Wing
operations during Allied Force.
Beyond this direct research support, I am grateful to Brigadier Gen-

eral Dan Darnell, commander, 31st Fighter Wing, for having pro-
vided me an opportunity to fly on a Block 40 F-16CG laser-guided
bomb delivery training mission out of Aviano to sample at first hand
Acknowledgments xxvii
the challenge of 100-percent target identification from medium alti-

tude, as well as other operational problems associated with Allied
Force; Lieutenant General Wald for having arranged a Block 50
F-16CJ SEAD mission orientation for me with the 20th Fighter Wing
at Shaw AFB, South Carolina; General Martin, USAFE’s commander,
for having allowed me to fly on an F-15E Kosovo-type strike training
mission into the Scottish highlands with the 48th Fighter Wing, RAF
Lakenheath, England; and Lieutenant General Fred McCorkle,
USMC, deputy commandant for aviation, Hq United States Marine
Corps, for his support in enabling me to get a first-hand exposure to
Marine F/A-18D operations during two ground-support training sor-
ties with Marine All-Weather Fighter/Attack Squadron 332 at MCAS
Beaufort, South Carolina. This hands-on familiarization with some
of the principal mission employment practices used in Operation Al-
lied Force was uniquely helpful in informing my characterization of
those practices in this assessment.
Finally, for their helpful comments on all or parts of an earlier draft, I

acknowledge General W. L. Creech, USAF (Ret.); General Jumper; Air
Chief Marshal Day; Lieutenant General Short; Lieutenant General
Bruce Carlson, USAF, director, J-8, Joint Staff; Lieutenant General
Ronald Keys, commander, 16th Air Force; Lieutenant General John
Dallager, former DCS/Operations, SHAPE, and now superintendent,
U.S. Air Force Academy; Air Vice Marshal Tony Mason, RAF (Ret.);
Lieutenant General (select) Lance Smith, commander, and Colonel
Steve Carey, Lieutenant Colonel Russ Barnes, and Major Richard
Leatherman, Hq Air Force Doctrine Center; Major General David
Deptula, director, USAF Quadrennial Defense Review; Major General
(select) Leaf; Brigadier General Robert Bishop, deputy director of
operations, Hq United States Air Force; Brigadier General Darnell,
Colonel Eberhart, and Lieutenant Colonel Dissinger of the 31st
Fighter Wing at Aviano; Colonel Charles Westenhoff, USAF, director,
Joint SEAD Joint Test Force; Colonel Wesley Jarmulowicz, USMC,
formerly assigned to AFSOUTH J-5 and now commander, Marine
Aircraft Group 31; Colonel Robert Owen, USAF, Hq Air Mobility
Command; Professor David Mets, School of Advanced Airpower
Studies, Air University; Robert Haffa and Barry Watts, Northrop
Grumman Analysis Center; Mark Butler and Daniel Harrington, Hq
USAFE; Bob Johnston and Major Mike Pietrucha, Hq USAF; and my
RAND colleagues Nora Bensahel, Tom Hamilton, Ted Harshberger,
xxviii NATO’s Air War for Kosovo: A Strategic and Operational Assessment
Steve Hosmer, Dave Kassing, Rob Mullins, Bruce Nardulli, David

Ochmanek, Bruce Pirnie, Carl Rhodes, John Stillion, and Alan Vick. I
am additionally indebted to Emily Rogers for her deft assistance with
composition and formatting and to Miriam Polon for her invariably
sure editing. My special thanks go to Ivo Daalder of the Brookings
Institution and Alan Gropman of the Industrial College of the Armed
Forces for their careful and constructive reviews of the final product.
ACRONYMS
AAA Antiaircraft Artillery

AB Air Base
ABCCC Airborne Command and Control Center
ACC Air Combat Command
ACTORD Activation Order
AEF Air Expeditionary Force
AEW Airborne Early Warning
AFB Air Force Base
AFSOUTH Allied Forces Southern Europe
AFV Armored Fighting Vehicle
AGM Air-to-Ground Missile
AMRAAM Advanced Medium-Range Air-to-Air Missile
AOR Area of Responsibility
APC Armored Personnel Carrier
ATACMS Army Tactical Missile System
ATC Air Traffic Control
ATO Air Tasking Order
AWACS Airborne Warning and Control System
AWOS Air War Over Serbia
xxix
xxx NATO’s Air War for Kosovo: A Strategic and Operational Assessment
BDA Battle Damage Assessment

BDI Battle Damage Indications
C2 Command and Control
CALCM Conventional Air-Launched Cruise Missile
CAOC Combined Air Operations Center
CAS Close Air Support
CBS Columbia Broadcasting System
CBU Cluster Bomb Unit
CIA Central Intelligence Agency
CINC Commander in Chief
CINCCENT CINC U.S. Central Command
CINCEUR CINC U.S. European Command
CINCPAC CINC U.S. Pacific Command
CINCSOUTH CINC Allied Forces Southern Europe
CNN Cable News Network
CNO Chief of Naval Operations
COMAIRCENT Commander Allied Air Forces Central Europe
COMAIRSOUTH Commander Allied Air Forces Southern
Europe
COMUSAFE Commander United States Air Forces in
Europe
CONOPLAN Concept of Operations Plan
CSAR Combat Search and Rescue
DEAD Destruction of Enemy Air Defenses
DMPI Desired Mean Point of Impact
DMSP Defense Meteorological Support Program
DSCS Defense Satellite Communications System
Acronyms xxxi
DSP Defense Support Program

ECM Electronic Countermeasures
ECR Electronic Combat Role
ELINT Electronic Intelligence
EO Electro-Optical
EW Early Warning
EW Electronic Warfare
FAC Forward Air Controller
FLIR Forward-Looking Infrared
FY Fiscal Year
GAT Guidance, Apportionment, and Targeting
GATS GPS-Aided Targeting System
GBU Glide Bomb Unit
GPS Global Positioning System
HARM High-Speed Antiradiation Missile
HMS Her Majesty’s Ship
HTS HARM Targeting System
HUD Head-Up Display
IADS Integrated Air Defense System
ID Identification
IFF Identification Friend or Foe
IFOR NATO Implementation Force in Bosnia and
Herzegovina
IMF International Monetary Fund
INS Inertial Navigation System
IR Infrared
ISR Intelligence, Surveillance, and Reconnaissance
xxxii NATO’s Air War for Kosovo: A Strategic and Operational Assessment
JAC Joint Analysis Center

JCS Joint Chiefs of Staff
JDAM Joint Direct Attack Munition
JFACC Joint Force Air Component Commander
JTF Joint Task Force
JTIDS Joint Tactical Information Distribution System
JWAC Joint Warfare Analysis Center
KEZ Kosovo Engagement Zone
KFOR Kosovo Force
KLA Kosovo Liberation Army
LANTIRN Low-Altitude Navigation and Targeting
Infrared for Night
LD/HD Low Density/High Demand
LGB Laser-Guided Bomb
LOC Line of Communication
MANPADS Man-Portable Air Defense System
MEU Marine Expeditionary Unit
MFD Multifunction Display
MLRS Multiple-Launch Rocket System
MSTS Multisource Tactical System
MTI Moving Target Indicator
MTW Major Theater War
MUP Serbian Interior Ministry Police
NAC North Atlantic Council
NATO North Atlantic Treaty Organization
NCA National Command Authorities
NIMA National Imagery and Mapping Agency
Acronyms xxxiii
NPIC National Photographic Interpretation Center

NRO National Reconnaissance Office
OCA Offensive Counterair
ORI Operational Readiness Inspection
OSCE Organization for Security and Cooperation in
Europe
PACAF Pacific Air Forces
PGM Precision-Guided Munition
POL Petroleum, Oil, and Lubricants
POW Prisoner of War
RAF Royal Air Force
RCS Radar Cross-Section
RNLAF Royal Netherlands Air Force
ROE Rules of Engagement
RWR Radar Warning Receiver
SACEUR Supreme Allied Commander Europe
SAM Surface-to-Air Missile
SAR Synthetic Aperture Radar
SAS Special Air Service
SEAD Suppression of Enemy Air Defenses
SFOR NATO Stabilization Force in Bosnia and
Herzegovina
SHAPE Supreme Headquarters Allied Powers Europe
SIGINT Signals Intelligence
STARS Surveillance Target Attack Radar System
TACAN Tactical Air Navigation
TARPS Tactical Air Reconnaissance Pod System
xxxiv NATO’s Air War for Kosovo: A Strategic and Operational Assessment
TF Task Force
TIP Tactical Integrated Planning
TLAM Tomahawk Land-Attack Missile
TOT Time on Target
UAV Unmanned Aerial Vehicle
UCAV Unmanned Combat Air Vehicle
UHF Ultra-High Frequency
UN United Nations
USA United States Army
USAF United States Air Force
USAFE United States Air Forces in Europe
USAFE/SA United States Air Forces in Europe, Studies
and Analysis Office
USAREUR United States Army in Europe
USEUCOM United States European Command
USMC United States Marine Corps
USN United States Navy
USS United States Ship
VJ Yugoslav Army
VTC Video Teleconference
Chapter One
INTRODUCTION
Between March 24 and June 9, 1999, NATO, led by the United States,
conducted an air war against Yugoslavia in an effort to halt and re-
verse the continuing human-rights abuses that were being commit-
ted against the citizens of its Kosovo province (see the Frontispiece,
Map of Kosovo) by Yugoslavia’s elected president, Slobodan Milose-
vic. As it turned out, that 78-day effort, called Operation Allied Force,
represented the third time in a row during the 1990s, after Opera-
tions Desert Storm and Deliberate Force, in which air power proved
pivotal in determining the outcome of a regional conflict. Yet
notwithstanding its ultimate success, what began as a hopeful gambit
for producing quick compliance on Milosevic’s part soon devolved,
for a time at least, into a seemingly ineffectual bombing experiment
with no clear end in sight. Not only was the operation’s execution
hampered by uncooperative weather and a surprisingly resilient
opponent, it was further afflicted by persistent hesitancy on the part
of U.S. and NATO decisionmakers that was prompted by fears of
inadvertently killing civilians and losing friendly aircrews, as well as
by sharp differences of opinion within the most senior U.S.
command element over the best way of applying allied air power
against Serb assets to achieve the desired effects. All of that and
more, however unavoidable some aspects of it may have been, made
NATO’s air war for Kosovo a substantial step backward in efficiency
when compared to Desert Storm.
This book assesses Operation Allied Force from a strategic and op-
erational perspective, with a view toward spotlighting what was most
gratifying about the application of allied air power throughout the ef-
fort, as well as identifying and exploring aspects of air power’s per-
1
2 NATO’s Air War for Kosovo: A Strategic and Operational Assessment
formance that indicated continued deficiencies in need of attention.

The analysis is based entirely on openly accessible information, en-
riched at various points by inputs gleaned from interviews with se-
lected Allied Force participants at both the command and execution
levels. Although the U.S. government has yet to release many of the
more recondite statistics associated with the air war’s prosecution at
the operational and tactical levels, more than enough confirmed in-
formation on the broader essentials has now been made public by
the Department of Defense and by leading NATO officials to permit a
confident reconstruction of what happened during Operation Allied
Force. As in the case of the Persian Gulf War a decade ago, the prin-
cipal distinguishing features of NATO’s air war for Kosovo are no
longer in dispute. What remains in contention are their meaning
and implications.
As more details about the background and conduct of the Kosovo air
war have become available, the debate over Allied Force and over the
appropriate “lessons” to be drawn from it has tended to fragment
into what one observer called “a series of mini-arguments about de-
tails and facts and figures,” perhaps most notoriously with respect to
how many enemy tanks were destroyed by allied air attacks and
whether U.S. and NATO officials conspired to cover up the surpris-
ingly poor performance of the air effort in that respect. 1 In contrast,
this study seeks to maintain a steady lock on the larger picture.
Although it freely ventures into operational and technical detail
wherever appropriate, including on the tank issue and on related
questions concerning how various items of equipment worked, it fo-
cuses more on such broader questions as what U.S. and allied air
power accomplished by way of achieving their goals and what the
operation’s experience revealed about the state of air power’s con-
tinuing evolution as an instrument of joint and combined warfare.
Toward that end, the book first describes the air war’s strategic and
operational highlights in chronological order. It then considers the
various factors that interacted to induce Milosevic to capitulate when
he did. After that, it explores air power’s principal accomplishments,
as well as the many problems that worked to render Allied Force a
______________
1Christopher Cviic, “A Victory All the Same,” Survival, Summer 2000, p. 174.
Introduction 3
less than uniformly satisfactory performance by allied air power.2

The final chapter reviews Operation Allied Force in political and
strategic context and reflects on the most policy-pertinent conclu-
sions to be drawn from the experience. Because of the study’s pre-
dominant focus on matters pertaining to planning and execution, it
does not consider, other than by way of brief stage-setting, the poli-
tics and diplomacy that immediately preceded the air war, let alone
the deeper historical roots of the crisis.
______________
2 In the latter respect, this assessment consciously seeks to avoid the common
syndrome of so-called lessons-learned efforts whereby “losers tend to study what went
wrong while winners study what went right.” Princeton University political scientist
Bernard Lewis, an adviser to the USAF’s Gulf War Air Power Survey conducted in
1991–1992, called this cautionary reminder to the attention of the survey team as that
effort was getting under way. Quoted in Gian P. Gentile, How Effective Is Strategic
Bombing? Lessons Learned from World War II to Kosovo, New York, New York
University Press, 2001, p. 182.
This page intentionally blank
Chapter Two
PRELUDE TO COMBAT
Operation Allied Force, largely prompted by humanitarian concerns,

was a response by the United States and NATO to the steadily
mounting Serb atrocities that were being committed against the
ethnic Albanians who made up the vast majority of Kosovo’s
population. At bottom, the crisis was rooted in a centuries-long
history of Balkan strife.1 Its more immediate origins could be traced
back to a decade before, when the disintegration of the Yugoslav
______________
1 For informed insight into the origins of the ancestral hatreds that animated the
atrocities committed against the Kosovar Albanians by Serbia, one can do no better
than the epic novel by Ivo Andric, The Bridge on the Drina, Chicago, University of
Chicago Press, 1977. Written in Serbo-Croatian in 1945, this tour de force won the
1961 Nobel prize for literature. It speaks about Balkan conflicts from the earliest
clashes between the Bosnian Turks and Serb Christians in the early 15th century to the
coming of the First World War. In a passage hauntingly reminiscent of more recent
Balkan horrors, Andric described how ethnic rivals as far back as the 17th century
“were as if drunk with bitterness, from desire for vengeance, and longed to punish and
kill whomsoever they could, since they could not punish or kill those whom they
wished” (pp. 86–87). Of a later generation looking at the redrawn map of Bosnia after
the Balkan war of 1912, Andric likewise wrote that they “saw nothing in those curving
lines, but they knew and understood everything, for their geography was in their blood
and they felt biologically their picture of the world” (p. 229). A balanced synopsis of
this history that places it in the context of the 20th-century developments that led up
to the 1999 Kosovo crisis is presented in William W. Hagen, “The Balkans’ Lethal Na-
tionalisms,” Foreign Affairs, July/August 1999, pp. 52–64. For more on this back -
ground, see Ivo H. Daalder and Michael E. O’Hanlon, Winning Ugly: NATO’s War to
Save Kosovo, Washington, D.C., Brookings Institution, 2000, pp. 1–100. See also
Michael Ignatieff, Virtual War: Kosovo and Beyond, New York, Metropolitan Books,
2000, especially pp. 11–65; Misha Glenny, The Balkans: Nationalism, War and the
Great Powers, 1809–1999, New York, Penguin Books, 2000; and Tim Judah, Kosovo:
War and Revenge, New Haven, Connecticut, Yale University Press, 2000.
5
Federation began during the waning years of the cold war. 2 Under
the iron rule of Yugoslavia’s independent communist leader, Marshal
Josip Broz Tito, Kosovo had remained an autonomous and self-
governing province of Serbia for nearly 40 years, and members of its
largely ethnic Albanian populace were able to live a reasonable
approximation of normal lives. Once communist rule began to un-
ravel in the late 1980s after Tito’s death, however, the Serb minority
in Kosovo reacted forcefully against what they perceived to be willful
discrimination against them by the Kosovar Albanian authorities.
After winning an election in 1989 in which he played heavily on Serb

feelings of mistreatment, Milosevic decreed an end to Kosovo’s au-
tonomy, imposed Serb rule, and unleashed a resurgence of ethnic
violence throughout the former Yugoslav Federation that, by 1995,
had caused more than a quarter of a million innocent citizens to lose
their lives in a renewed Balkan civil war. That, in turn, gave rise to a
group of Kosovar Albanian nationalists who espoused the use of vio-
lence in pursuit of Kosovar independence, ultimately spawning a
militant émigré group called the Kosovo Liberation Army (KLA, or
UCK in Albanian, for Ushtria Clirimtare e Kosoves), whose members
began waging a partisan war against the Serb army and police units
that controlled the increasingly conflicted province.
In February 1998, in an escalating wave of reprisals against the rear-

guard actions of this nascent band of ethnic Albanian guerrillas, a
unit of the Yugoslav interior ministry police, or MUP (for Minister-
stvo Unuprasnij Poslava), counterattacked in force against the KLA in
the Drenica region of Kosovo, wantonly killing some 80 Kosovar Al-
banian civilians in the process. In response, U.S. special envoy
Richard Holbrooke was sent to Belgrade by the Clinton administra-
tion to beseech Milosevic to desist from further acts of violence
against Kosovar civilians. Milosevic refused. Later, in early fall of
1998, some 30,000 Kosovar civilians were forced to flee their homes
in the wake of resurgent Serb pillaging and terrorizing of the Kosovo
countryside. That renewed violence prompted the passage on
September 23, 1998, of UN Security Council Resolution 1199 warning
of an “impending humanitarian catastrophe” and calling for an im-
______________
2Today, Serbia and Montenegro (the latter is semiautonomous) are all that remain of
the former Yugoslavia.
Prelude to Combat 7
mediate halt to the escalating strife in Kosovo. The following month,

the Clinton administration again dispatched Holbrooke to Belgrade
in a bid to persuade Milosevic to agree to negotiations on Kosovar
autonomy and to accept a presence in Kosovo of unarmed interna-
tional monitors from the Organization for Security and Cooperation
in Europe (OSCE) to verify Serb compliance with Resolution 1199.
In the end, Milosevic assented to negotiations and agreed to permit

an OSCE verification mission to enter Kosovo after the endorsement
by the North Atlantic Council (NAC), the political arm of NATO, of an
Activation Order (ACTORD) laying the groundwork for NATO air at-
tacks against Serb military targets as an inducement. The mission,
headed by U.S. diplomat William Walker, aimed at ensuring that the
KLA’s partisans remained in the mountains and that the Yugoslav
army, or VJ (for Vojska Jugoslavskaya), remained in its garrisons. Air
surveillance for this OSCE presence was to be provided by NATO,
and a NATO extraction force began forming in Macedonia to with-
draw the OSCE observers under armed protection in case events
turned sour enough to endanger their safety.
Before long, however, the OSCE monitors found themselves watch-

ing helplessly as the Serb killing of Kosovar Albanians continued at a
slow but steady rate. In response, NATO declared that it would take
“all steps,” including air strikes if necessary, to compel Serb compli-
ance in bringing about a settlement in Kosovo. Meantime, as Hol-
brooke continued shuttling between Belgrade and Washington,
Kosovo remained all but absent from the Clinton administration’s
list of priorities. Among other preoccupations closer to home, final
preparations for the president’s impeachment trial in the House of
Representatives had entered full swing by December 1998, and ten-
sions with Iraq were about to escalate into the launching of Opera-
tion Desert Fox that same month, a four-day mini-air operation
waged by U.S. and British forces in what turned out to have been an
almost entirely symbolic and ineffectual response to Saddam Hus-
sein’s earlier summary decision to refuse further cooperation with
UN arms inspectors.
The trigger event that finally spurred the Clinton administration into
action with respect to Kosovo occurred on January 15, 1999, when
MUP and Serb paramilitary troops in hot pursuit of KLA fighters en-
tered the village of Racak and proceeded to slaughter 45 hapless
ethnic Albanian civilians. Ambassador Walker personally traveled to
Racak the next day to view the carnage, calling it a crime against hu-
manity and all but blaming Milosevic by name for having ordered it.
Two days later, the Yugoslav government, in response, declared
Walker persona non grata and issued an expulsion order, which
Walker ignored.3 The Racak massacre signaled the beginning of the
end to any further active role for the OSCE monitors, who were now
increasingly at physical risk themselves and who were ultimately
withdrawn less than a week before the commencement of Operation
Allied Force. It turned out to be a serious miscalculation on Bel-
grade’s part. What Milosevic may have thought was “just another
village” proved to be one too many as far as the United States and
NATO were concerned.4 The event marked the beginning of the final
countdown toward NATO’s ultimate decision to proceed with Allied
Force. On January 30, the NAC approved the launching of NATO air
attacks against Serbia if the Serb leaders continued to refuse nego-
tiations with their Kosovar counterparts.
On February 6, the Contact Group (made up of representatives from

France, Germany, Great Britain, Russia, and the United States),
prodded by Secretary of State Madeleine Albright, summoned Serb
and KLA representatives to the Rambouillet chateau on the outskirts
of Paris for a last-chance round of talks aimed at producing an over-
arching settlement for Kosovo. Those talks ended without agree-
ment on February 23. Further talks began in Paris on March 15.
During the latter negotiations, Albright delivered an ultimatum to the
Serbs and Kosovars alike that, as an incentive, offered to contribute
28,000 NATO peacekeepers, including 4,000 U.S. troops, to police any
negotiated settlement. Three days later, the KLA signed a peace ac-
cord aimed at giving Kosovo broad autonomy within Serbia. The day
after, however, Serbia refused to sign, insisting that it would not even
consider the idea of foreign troops on Kosovo soil. More ominously
yet, on the very same day that this second round of talks began, Milo-
______________
3The expulsion order was later rescinded by Milosevic.
4On this point, NATO’s Secretary General, Javier Solana, remarked that a Serb diplo-
mat had been heard to cite a rule of thumb to the effect that “a village a day would
keep NATO away.”
Prelude to Combat 9
sevic ordered a major escalation of the buildup of VJ forces both

within and immediately adjacent to Kosovo that had begun the pre-
vious month, in a clear sign that a major move against the KLA and
against ethnic Albanian civilians was imminent.
By all indications, Milosevic did not enter the Rambouillet process

with any intent to negotiate seriously. On the contrary, in all likeli-
hood he saw it as presenting a perfectly timed opportunity to posi-
tion himself to launch Operation Horseshoe (Potkova), as his incipi-
ent ethnic cleansing campaign was code-named. By that point, he
most likely fully anticipated that NATO would eventually bomb him,
much as U.S. and British forces did in a token manner against Iraq
two months previously in Desert Fox. Probably key to Milosevic’s
strategy was an underlying belief that he could take at least as much
measured pain from a symbolic NATO air operation as Saddam Hus-
sein had endured from Desert Fox. That belief most likely hinged on
an associated conviction that NATO’s limited tolerance for bombing
would run out in short order and that he would then have Kosovo all
to himself, with no further outside meddling by NATO, OSCE, or any
other foreign peace-enforcement entities, and with no Kosovar Al-
banians. As if to bear that out, not only did Belgrade reject NATO’s
peace proposal outright, it simultaneously launched a new campaign
of burning and pillaging by some 40,000 VJ troops in the central
Drenica region of Kosovo, using tanks, heavy artillery, and mortar
fire against dozens of villages. In the process, VJ and MUP forces
destroyed three of seven known regional headquarters of the KLA
and forced thousands of ethnic Albanian civilians to flee. In the wake
of that renewed assault, the UN High Commissioner for Refugees re-
ported 240,000 internally displaced persons in Kosovo, including
60,000 rendered homeless in just the preceding three weeks.
The refugee crisis quickly assumed all the earmarks of a humanitar-

ian disaster. President Clinton ordered Holbrooke back to Belgrade
on March 22 in an eleventh-hour bid to persuade Milosevic to desist
from further ravaging of Kosovo or else face NATO bombing attacks.
Holbrooke was instructed to warn Milosevic that NATO was prepar-
ing air and missile strikes that would destroy much of Yugoslavia’s
military infrastructure. Milosevic was further warned that the targets
of those attacks would be not just in Kosovo but in Serbia as well.5

Holbrooke made no attempt to bargain and stressed to Milosevic
that he was in Belgrade solely to deliver a message. At the end of a
four-hour meeting, he was rebuffed.
At that point, with the gauntlet thrown down by Holbrooke, U.S. of-
ficials presented NATO’s ambassadors with a final proposed bomb-
ing plan against Serbia, the declared goals of which were a verifiable
halt to ethnic cleansing and atrocities on the ground in Kosovo; a
withdrawal of all but a token number of VJ, MUP, and paramilitary
troops from Kosovo; the deployment of an international peacekeep-
ing force in Kosovo; the return of refugees and their unhindered ac-
cess to aid; and the laying of groundwork for a future settlement in
Kosovo along the lines of the Rambouillet terms of reference.6
Commenting on the threatened campaign, the Supreme Allied
Commander in Europe (SACEUR), U.S. Army General Wesley Clark,
warned that “if required, we will strike in a swift and severe fashion.”
General Klaus Naumann, the chairman of NATO’s Military Commit-
tee, added that Milosevic was “severely mistaken” if he believed that
NATO would engage merely in pinprick attacks and then await his
response. 7
Earlier, when the allies had empowered NATO Secretary General

Javier Solana to authorize air strikes on January 30, the declared in-
tent was to conduct limited raids over 48 hours and then pause to
encourage Milosevic to reconsider. The plan this time was for a
wider range of targets to be hit and for a longer operation aimed at
causing considerable infrastructure damage. In an eleventh-hour
bid to marshal public support for the impending air effort, President
Clinton made an appeal, in a televised speech at a labor union lun-
______________
5 In his exchange with Milosevic, Holbrooke said: “You understand our position?”
Milosevic: “Yes.” Holbrooke: “Is it absolutely clear what will happen when we leave,
given your position?” Milosevic: “Yes, you will bomb us. You are a big and powerful
nation. You can bomb us if you wish.” Bruce W. Nelan, “Into the Fire,” Time, April 5,
1999, p. 35. Later, Holbrooke added that Milosevic was “tricky, evasive, smart, and
dangerous,” further noting that his mood in the final confrontation was “calm, almost
fatalistic, unyielding.” “‘He Was Calm, Unyielding,’” Newsweek, April 5, 1999, p. 37.
6Jane Perlez, “Holbrooke to Meet Milosevic in Final Peace Effort,” New York Times,
March 22, 1999.
7R. Jeffrey Smith, “Belgrade Rebuffs Final U.S. Warning,” Washington Post, March 23,
1999.
Prelude to Combat 11
cheon on the day before the bombing commenced, beseeching the

American people to support his actions in coming to grips with
NATO’s looming Kosovo predicament.8
To be sure, planning for an air operation of some sort against Serbia

had begun as early as June 1998. Initial plans were for an option
called Operation Nimble Lion, which would have pitted a substantial
number of U.S. and allied aircraft against some 250 targets through-
out the former Yugoslavia.9 This option was developed wholly within
U.S. channels by the 32nd Air Operations Group at Ramstein Air
Base, Germany, at the behest of USAF General John Jumper in his
capacity as commander, United States Air Forces in Europe (USAFE),
in response to a directive from Clark in his capacity as commander in
chief, U.S. European Command (USEUCOM). A separate plan called
Concept of Operations Plan (CONOPLAN) 10601 was later developed
by NATO and approved by the NAC. Although there was some over-
lap between these two plans, the thrust of each was different. Nimble
Lion would have hit the Serbs hard at the beginning, whereas 10601
entailed a gradual, incremental, and phased approach. The latter
ultimately became the basis for Operation Allied Force.
Two closely related U.S. joint task force (JTF) planning efforts called
Operations Flexible Anvil (commanded by U.S. Navy Vice Admiral
Daniel Murphy, commander of the 6th Fleet) and Sky Anvil
(commanded by USAF Lieutenant General Michael Short, comman-
der of the 16th Air Force at Aviano Air Base, Italy) followed in the
summer of 1998.10 Those efforts were terminated when Milosevic
initially agreed to a cease-fire after his October 5–13 talks with
______________
8Charles Babington and Helen Dewar, “President Pleads for Support,” Washington
Post, March 24, 1999.
9Telephone conversation with Lieutenant General Michael Short, USAF (Ret.), August
22, 2001.
10Flexible Anvil was a U.S.-only option that envisaged only ship-launched Tomahawk
and conventional air-launched cruise missile attacks over a 48- to 72-hour period,
roughly along the lines of Operation Desert Fox conducted against Iraq the following
December. Sky Anvil envisaged follow-on air strikes in a transition to a NATO opera-
tion (or an operation involving a more truncated coalition of the willing). General
Short believed that it was counterproductive to fragment these closely connected op-
tions into two separate plans, but he and Admiral Murphy were well acquainted and
kept each other informed. Conversation with Vice Admiral Daniel J. Murphy, USN,
commander, 6th Fleet, aboard the USS LaSalle, Gaeta, Italy, June 8, 2000.
Holbrooke.11 In all, General Jumper later reported that by the onset

of Allied Force, no fewer than 40 air campaign options had been
generated and fine-tuned. 12 Those options were said to have
included some that were at least implicitly critical of the proposed
use of NATO air power without a supporting ground threat to
encourage enemy troops to assemble and maneuver so they might
be more easily targeted and attacked from the air.
In the end, however, the plan ultimately agreed to by NATO expressly

ruled out any backstopping by ground forces for two avowed rea-
sons. The first had to do with identified logistic difficulties, the antic-
ipated challenge of the terrain, and poor access and basing oppor-
tunities. The second, and far more pivotal, reason entailed the
Clinton administration’s concern over lack of congressional support
for such an option and the presumed unwillingness on the part of the
American people and the NATO allies to accept combat casualties,
reinforced by a near-certainty that the allies would not buy into a
ground option. All planning, moreover, took for granted that NATO’s
most vulnerable area (or “center of gravity”) was its continued cohe-
sion as an alliance. In light of that, any target or attack tactic deemed
even remotely likely to undermine that cohesion, such as the loss of
friendly aircrews, excessive noncombatant casualties, excess collat-
eral damage to civilian structures, or anything else that might un-
dermine domestic political support or cause a withdrawal of public
backing for the bombing effort, was to be most carefully consid-
ered—if not avoided altogether.
______________
11Lieutenant Colonel L. T. Wight, USAF, “What a Tangled Web We Wove: An After-
Action Assessment of Operation Allied Force’s Command and Control Structure and
Processes,” unpublished paper, p. 1.
12General John Jumper, USAF, testimony to the Military Readiness Subcommittee,
House Armed Services Committee, Washington, D.C., October 26, 1999. The most
fully developed of these iterations, called Operation Allied Talon, was a true phased air
campaign plan rooted in effects-based targeting and aimed at achieving concrete mili-
tary objectives. Despite the best efforts of the JTF Noble Anvil leadership (Admiral
James Ellis, General Jumper, and General Short) to sell this plan to SACEUR, General
Clark never adopted it. Instead, he elected to cut and paste different elements of the
different plans that he thought were most appropriate and labeled the resultant
product Operation Allied Force. Comments on an earlier draft by Hq USAFE/SA, April
6, 2001. General Jumper himself later confirmed that Allied Talon was a nonstarter.
Conversation with General John P. Jumper, USAF, Hq Air Combat Command, Langley
AFB, Virginia, May 15, 2001.
NATO’s final plan was conceived from the start as a coercive opera-
tion only, with the implied goal of inflicting merely enough pain to
persuade Milosevic to capitulate. Its first phase, against only 51 ap-
proved integrated air defense system (IADS) targets and 40 approved
punishment targets out of 169 in NATO’s Master Target File, entailed
attacks against a combination of enemy air defenses and fixed army
installations that aimed at softening up Yugoslavia’s IADS and
demonstrating NATO’s ability to conduct precise air attacks with a
minimum of unintended damage. The second phase envisaged at-
tacks against military targets mainly, though not exclusively, below
the 44th parallel, which bisected Yugoslavia well south of Belgrade
(see Figure 2.1). Only in the third phase, if need be, would the
RAND MR1365-2.1
AUSTRIA
HUNGARY
ROMANIA
Taszar
Aviano SLOVENIA
Ljubljana Novi Sad
Vicenza Zagreb CROATIA Danube
River
Belgrade
BOSNIA AND
HERZEGOVINA arallel
Cervia 44th p
BULGARIA
Sarajevo SERBIA
Nis
Sofia
KOSOVO
Amendola MONTENEGRO
Pristina
Podgorica
ADRIATIC
SEA Kukes Skopje
Rome MACEDONIA
ITALY Durres
Tirana
ALBANIA
Naples
Gioia del Colle Brindisi
GREECE
Figure 2.1—Allied Force Area of Operations

bombing go in earnest after military facilities north of the 44th paral-

lel and against targets in Belgrade itself.13 NATO had approved this
three-phase plan in principle the preceding October as a part of its
ACTORD and had handed the keys for Phase I to Solana on January
30. Approval by the NAC of Phases II and III, however, would come
only after the air effort began.
For his part, General Clark had called for punitive air strikes against
Yugoslavia as early as January 1999, in response to the Serb massacre
of 45 Kosovar Albanians near the town of Racak just days before.
Persistent pressures from within NATO to explore a diplomatic solu-
tion, however, outweighed that recommendation for the early use of
force. The resulting delay gave Milosevic time to bolster his forces,
disperse important military assets, hunker down for an eventual
bombing campaign, and lay the final groundwork for the ethnic
cleansing of Kosovo. Owing to that delay, NATO lost any element of
surprise that may otherwise have been available.14
In the end, Operation Allied Force came just 10 days short of NATO’s
50th anniversary. The Clinton administration did not seek a UN
Security Council resolution approving the air attack plan, since it
knew that Russia and China had both vowed to veto any proposal
calling for air strikes.15 NATO’s going-in expectation was that the
bombing would be over very quickly. Indeed, so confident were its
______________
13Charles Babington and William Drozdiak, “Belgrade Faces the 11th Hour, Again,”
Washington Post, March 22, 1999. For more first-hand comment on the intra-NATO
politics that preceded Allied Force, see General Wesley K. Clark, Waging Modern War:
Bosnia, Kosovo, and the Future of Combat, New York, Public Affairs, 2001, especially
pp. 121–189.
14William Drozdiak, “Politics Hampered Warfare, Clark Says,” Washington Post, July
20, 1999.
15This should not be taken to suggest that NATO’s air war against Serbia was a uni-
lateral action undertaken without regard for the UN whatsoever. On the contrary, in
March 1998 the Security Council had expressly recognized in Resolution 1160 that the
Serb government’s repression of the ethnic Albanian population in Kosovo constituted
a threat to international peace and security, a view later repeated in Resolution 1199
six months before the start of Allied Force, which called for action aimed at heading off
“the impending humanitarian catastrophe” in Kosovo. As an IISS comment later
noted, NATO’s air war for Kosovo thus constituted “a highly significant precedent,” in
that it established “more firmly in international law the right to intervene on
humanitarian grounds, even without an express mandate from the Security Council.”
Strategic Survey 1999/2000, London, England, The International Institute for Strategic
Studies, 2000, p. 26.
principals that merely a token bombing effort would suffice to per-

suade Milosevic to yield that the initial attack was openly announced
in advance, with U.S. officials conceding up front that it would take a
day or more to program all of the TLAMs to hit some 60 planned aim
points.16 Only at the last minute did NATO’s political leaders give
Secretary General Solana authority for what one NATO official called
a “much more diverse target list, a more intensive pace of operations,
and an expanded geographical zone.”17 Once under way, the slowly
escalating air effort put the United States into two simultaneous re-
gional conflicts (the other being Operations Northern and Southern
Watch against Iraq) for the first time since World War II. It also made
for a uniquely demanding test for American air power and became
the most serious foreign policy crisis of the Clinton presidency.
______________
16Jane Perlez, “U.S. Option: Air Attacks May Prove Unpalatable,” New York Times,
March 23, 1999.
17Steven Erlanger, “U.S. Issues Appeal to Serbs to Halt Attack in Kosovo,” New York
Times, March 23, 1999.
Chapter Three
THE AIR WAR UNFOLDS
The operational setting of Yugoslavia contrasted sharply with the one

presented to coalition planners by Iraq in 1991. Defined by a series
of interwoven valleys partly surrounded by mountains and protected
by low cloud cover and fog, Serbia and Kosovo made up an arena
smaller than the state of Kentucky (39,000 square miles), with Kosovo
itself no larger than the Los Angeles metropolitan area. Its topogra-
phy and weather—compounded by an enemy IADS that was guaran-
teed to make offensive operations both difficult and dangerous—
promised to provide a unique challenge for NATO air power.
Yugoslavia’s air defenses were dominated by surface-to-air missile

(SAM) batteries equipped with thousands of Soviet-made SAMs, in-
cluding three SA-2 battalions; 16 SA-3 battalions, each with numer-
ous launchers directed by LOW BLOW fire-control radars; and five
SA-6 regiments fielding five batteries each, for a total of 25 SA-6 bat-
teries directed by STRAIGHT FLUSH radars. These radar-guided
SAMs were supplemented by around 100 vehicle-mounted SA-9 and
several SA-13 infrared SAMs, along with a profusion of man-portable
infrared SAMs, some 1,850 antiaircraft artillery (AAA) pieces, and nu-
merous stockpiled reserve weapons and buried communications
lines. Backing up these defenses, the Yugoslav air force consisted of
238 combat aircraft, including 15 MiG-29 and 64 MiG-21 fighter-
interceptors.1 Although the Yugoslav IADS employed equipment
and technologies that dated as far back as the 1960s, albeit presum-
______________
1“AWOS [Air War Over Serbia] Fact Sheet,” Hq USAFE/SA, December 17, 1999. See
also The Military Balance, 1998/99, London, International Institute for Strategic Stud-
ies, 1998, p. 100.
17
ably with selected upgrades, its operators knew U.S. tactics well and
had practiced air defense drills and honed their operational tech-
niques for more than four decades. They also had the benefit of
more equipment and better training than did the Bosnian Serbs in
1995. Finally, they enjoyed the advantage of being protected both by
mountainous terrain and by the cover of inclement weather when
the air war began.
In addition, Serbia’s SA-2s, SA-3s, and SA-6s were served by more

than 100 acquisition and tracking radars, all of which were internet-
ted by underground land lines and fiber optic cables. They were fur-
ther backstopped by a robust civilian and military visual observer
network that included covert Serb observers who monitored NATO
aircraft as they took off from their bases in Europe.2 In anticipation
of a possible air offensive, Yugoslav defense specialists had met the
month before in Baghdad with their Iraqi counterparts. Indeed, such
Yugoslav-Iraqi collaboration had long preceded the Kosovo crisis.
Baghdad had purchased some Yugoslav IADS equipment late during
the cold war before the onset of Desert Storm. Iraq also very likely
shared intelligence with Belgrade on U.S. suppression of enemy air
defenses (SEAD) tactics, as well as its own experience and recom-
mendations, in subsequent years.3 According to General Salko Be-
gic, the air commander for the Muslim-Croat federation in Bosnia
and a former service academy classmate of the Serb generals who
were running Yugoslavia’s air defenses when the air attacks began,
the intended tactic to be used against attacking NATO aircraft was to
create a killing zone below 10,000 ft by means of AAA, SA-7 infrared
SAMs, and Swedish Bofors man-portable air defenses.4
______________
2Discussions with former East European strategic and tactical SAM operators on IADS
visual observer employment doctrine, as reported to the author by Hq USAFE/IN, May
18, 2001.
3John Diamond, “Yugoslavia, Iraq Talked Air Defense Strategy,” Philadelphia Inquirer,
March 30, 1999.
4Michael R. Gordon, “NATO to Hit Serbs from 2 More Sides,” New York Times, May 11,
1999. This last system featured the Bofors 40mm gun tied to the Giraffe radar-based
low-altitude air defense system (LAADS). It was the only radar-cued (as opposed to
radar-directed) AAA weapon fielded in the war zone and possibly the most potent low-
altitude AAA threat because of its local Giraffe-based LAADS command and control
system. Peter Rackham, ed., Jane’s C4I Systems, 1994–95, London, Jane’s Information
Group, 1994, p. 107.
The Air War Unfolds 19
In commenting on this layered and redundant air defense net, USAF

chief of staff General Michael Ryan, who earlier had commanded
Operation Deliberate Force over Bosnia in 1995, frankly conceded in
congressional testimony before the start of the operation that “these
guys are very good” and that friendly aircraft and aircrew losses were
“a distinct possibility.”5 Ryan added that Yugoslavia’s IADS made for
a “very substantive air defense capability” and that the Serbs main-
tained a “very professional army and air defense corps.” Because of
the assessed robustness of the Yugoslav IADS, Pentagon planners
were said to have estimated before opening night that NATO could
lose as many as 10 aircraft in the initial wave of strikes.6
INITIAL ATTACKS AND THEIR EFFECTS

Operation Allied Force began against Yugoslavia on the night of
March 24, within minutes of President Clinton’s announcement that
air attacks were under way. The initial concept of operations envis-
aged night raids against so-called enabling targets, such as enemy air
defense assets, in order to create a more permissive operating envi-
ronment for subsequent attacks against other classes of targets. In
announcing the commencement of attacks, the president declared
that the operation had three goals: “To demonstrate the seriousness
of NATO’s opposition to aggression,” to deter Milosevic from “con-
tinuing and escalating his attacks on helpless civilians,” and, if need
be, “to damage Serbia’s capacity to wage war against Kosovo by seri-
ously diminishing its military capabilities.” At the same time, he
pointedly stressed: “I don’t intend to put our troops in Kosovo to
fight a war.”7 To those opening words, the chairman of the Joint
Chiefs of Staff, U.S. Army General Henry H. Shelton, added that
NATO would engage “the full range of his military capabilities” if
______________
5 Paul Richter, “U.S. Pilots Face Perilous Task, Pentagon Says,” Los Angeles Times,
March 20, 1999. In testimony to the Senate Armed Services Committee on the eve of
the war, Ryan added: “I ran the air campaign in Bosnia, and this defensive array is
much more substantive . . . two or three times more so. It is deep and redundant.
Those guys [in Bosnia] were good, but these guys are better. There is a very real possi-
bility we will lose aircraft trying to take it on.” David Atkinson, “Stealth Could Play Key
Role in Kosovo, Despite Bad Weather,” Defense Daily, March 23, 1999, p. 1.
6Bruce W. Nelan, “Into the Fire,” Time, April 5, 1999, p. 31.
7Francis X. Clines, “NATO Opens Broad Barrage Against Serbs as Clinton Denounces
‘Brutal Repression,’” New York Times, March 25, 1999.
Milosevic did not desist from his offensive in Kosovo.8 As noted ear-
lier, it was accepted as a given by the Clinton administration that
Milosevic would settle quickly. As Secretary of State Albright clearly
attested to this expectation in a television interview on the evening
that the air attacks began: “I don’t see this as a long-term opera-
tion.” 9
The air war commenced with 250 committed U.S. aircraft, including
120 land-based fighters, 7 B-52s, 6 B-2s, 10 reconnaissance aircraft,
10 combat search and rescue (CSAR) aircraft, 3 airborne command
and control center (ABCCC) aircraft, and around 40 tankers.10 As for
NATO’s additional 18 members, 13 contributed aircraft for use in the
operation, with 11 allies (Britain, France, Germany, Italy, Belgium,
Canada, Denmark, the Netherlands, Norway, Spain, and Turkey)
eventually participating in offensive and defensive air combat
operations of all types. The first wave of attacks on the night of
March 24 consisted of cruise missile launches only, featuring TLAMs
fired by four U.S. surface ships (including USS Gonzales and USS
Philippine Sea), two U.S. fast-attack submarines (USS Albuquerque
and USS Miami), and a British attack submarine (HMS Splendid)
operating in the Adriatic Sea. This initial wave further included
AGM-86C CALCMs launched against hardened enemy structures by
six B-52s flying outside Yugoslav airspace. The latter were the first
shots fired in the operation.11 The initial target hits occurred shortly
______________
8Paul Richter, “Time Is Not on the Side of U.S., Allies,” Los Angeles Times, March 25,
1999.
9John T. Correll, “Assumptions Fall in Kosovo,” Air Force Magazine, June 1999, p. 4.
10This study has taken special care to characterize Operation Allied Force as an “air
war” or an “air effort,” rather than as a full-fledged “air campaign.” Although that ef-
fort continues to be widely portrayed as the latter, formal Air Force doctrine defines an
air campaign as “a connected series of operations conducted by air forces to achieve
joint force objectives within a given time and area.” Air Force Basic Doctrine, Maxwell
AFB, Alabama, Hq Air Force Doctrine Center, AFDD-1, September 1997, p. 78. By that
standard, NATO’s air war for Kosovo did not attain to the level of a campaign, as did
the earlier Operations Desert Storm and Deliberate Force. Rather, it was a continu-
ously evolving coercive operation featuring piecemeal attacks against unsystemati-
cally approved targets, not an integrated effort aimed from the outset at achieving
predetermined and identifiable operational effects.
11The effectiveness of these initial standoff attacks was not impressive. During the
first two weeks, no B-52 succeeded in launching all eight of its CALCMs. In one in-
stance, six out of eight were said to have failed. Also, the two times that B-52s later
fired the AGM-142 Have Nap cruise missile, both launches were reportedly opera-
after 8 p.m. local time in the vicinity of Kosovo’s capital city of

Pristina, shutting down the electrical power grid and plunging the
city into darkness. The main commercial and military airfield at
nearby Batajnica was also hit. In all, 55 U.S. cruise missiles were
expended the first night.
These cruise-missile attacks were followed by fixed-wing air strikes

that continued throughout the night, primarily against air defense
targets such as SAM batteries and radar and military communica-
tions sites.12 Allied aircraft operated out of Italy, Spain, France,
Germany, the United Kingdom, and the Adriatic Sea.13 Their targets
included a radar site at Podgorica, the capital city of Montenegro. In
addition, NATO aircrews hit airfields in Serbia, Kosovo, and Mon-
tenegro, as well as electrical power generating facilities, weapons-
producing factories, military and police barracks, and command and
control nodes, including some aim points located north of Belgrade.
Among specific targets attacked were the VJ’s Kosovoski Junaci bar-
racks near Pristina in Kosovo, the Golobovci airport in Montenegro,
munitions stores at Danilovgrad, and other military targets at
Radovac, Sipcanik, and Ulcini.14 Allied pilots were instructed to take
no chances with enemy infrared SAMs and AAA and to honor an alti-
_____________________________________________________________
tional failures. See John D. Morrocco, David Fulghum, and Robert Wall, “Weather,
Weapons Dearth Slow NATO Strikes,” Aviation Week and Space Technology, April 5,
1999, p. 26, and William M. Arkin, “Kosovo Report Short on Weapons Performance
Details,” Defense Daily, February 10, 2000, p. 2.
12 An important qualification is warranted here. Although the opening-night ap-
proved aim points largely entailed fixed IADS targets, the limited attacks conducted
against them were not part of a phased campaign plan in which rolling back the en-
emy IADS was a priority. There was no strategic emphasis on IADS takedown in these
attacks. Comments on an earlier draft by Hq USAFE/IN, May 18, 2001.
13Italian bases used included Aviano, Gioio del Colle, Villafranca, Amendola, Cervia,
Gazzanise, Ghedi, Piacenza, Istrana, Falconara, Practica di Mare, Brindisi, and
Sigonella. German bases used were Royal Air Force (RAF) Bruggen, Rhein Main Air
Base (AB), Spangdahlem AB, and Ramstein AB. Bases made available by the United
Kingdom were RAF Fairford, RAF Lakenheath, and RAF Mildenhall. Spain provided
Moron AB, and France provided Istres. For a complete list of all participating allied air
assets, their units, and their bases, as well as a tabulation of the Yugoslav IADS and air
order of battle as of April 20, see Benoit Colin and Rene J. Francillon, “L’OTAN en
Guerre!” Air Fan, May 1999, pp. 12–19. See also John E. Peters, Stuart Johnson, Nora
Bensahel, Timothy Liston, and Traci Williams, European Contributions to Operation
Allied Force: Implications for Transatlantic Cooperation, Santa Monica, California,
RAND, MR-1391-AF, 2001.
14Robert Hewson, “Operation Allied Force: The First 30 Days,” World Air Power Jour-
nal, Fall 1999, p. 16.
tude floor (or “hard deck”) of 15,000 ft to remain above their killing
envelopes.
Targets attacked the first night were reviewed with special care at the
White House by President Clinton, Secretary of Defense William Co-
hen, and General Shelton. Some proposed targets were removed
from the list by dissenting NATO leaders out of concern for causing
collateral damage because of their close proximity to civilian build-
ings. In other borderline cases in which targets were reluctantly ap-
proved, the recommended bomb size was reduced to minimize or
preclude collateral damage. One of every five laser-guided bombs
dropped by an F-117 the first night was a 500-lb GBU-12 instead of a
2,000-lb GBU-27. That meant less likelihood of the bomb’s causing
inadvertent collateral damage, but also a lower probability of de-
stroying the intended target. The rules of engagement were un-
compromisingly restrictive, with pilots instructed to return home
with their weapons unless their assigned target could be positively
identified.15
In all, some 400 sorties were flown the first night, including 120 strike
missions against 40 targets consisting of five airfields, five army gar-
risons, communications centers, and storage depots, in addition to
IADS facilities. Only a few SA-3 and SA-6 SAMs were launched
against attacking NATO aircraft the first night. All the same, Pen-
tagon officials anticipated the day after that at least a dozen NATO
aircraft losses could be incurred should the operation continue be-
yond just a few days.16 Contrary to early Western press reports, Serb
IADS operators never intentionally husbanded their SAMs. Instead,
after experiencing allied SEAD operations for the first time, they
adapted their tactics to balance lethality with survivability, with the
result that they were always present and aggressive—even as they
showed greater firing discipline than the Iraqis did during Desert
Storm.17
______________
15Nelan, “Into the Fire,” p. 32.
16Steven Lee Myers, “Early Attacks Focus on Web of Air Defense,” New York Times,
March 25, 1999.
17Comments on an earlier draft by Hq USAFE/IN, May 18, 2001.
Numerous enemy fighters, including at least a dozen MiG-29s,

sought to engage attacking NATO aircraft the first night.18 One MiG-
29 was reported to have fired an R-73 (NATO code-named AA-10
Alamo) radar-guided missile toward an ingressing NATO fighter in an
ineffectual attempt to get off a counteroffensive shot. Two MiG-29s
were downed by USAF F-15s and one by a Dutch F-16. In addition, a
MiG-21 was believed to have crashed during an attempt to land.
Only rarely did Serb fighters rise to challenge NATO aircraft after
that. The following day, General Clark declared that the bombard-
ment would “systematically and progressively attack, disrupt, de-
grade, devastate,” and “ultimately . . . destroy” Milosevic’s army if he
failed to accept the American-drafted peace plan. Clark further de-
clared that the air effort would be “just as long and difficult as Presi-
dent Milosevic requires it to be.”19
Attacks carried out by NATO aircrews the second night were de-
scribed as “significantly heavier” than those the first night. Targets
included the VJ barracks at Urosevac and Prizren in Kosovo; the mili-
tary airfields at Nis in southern Serbia and Golubovci near Podgorica,
Montenegro; and other Serb military facilities near Trstenik and
Danilovgrad. 20 That night, fewer than 10 SAMs were fired, none of
which succeeded in scoring a hit. The third afternoon, a USAF F-15C
downed two more MiG-29s, which evidently had lost contact with
their ground controller and inadvertently strayed into Bosnian
airspace. Although their intended NATO targets were never posi-
tively determined, it was the subsequent conclusion of the allied air
commander, Lieutenant General Short, that the Serb pilots had sim-
ply lost any semblance of air situation awareness and, as a result, set
themselves up as easy prey for the F-15. 21
______________
18This suggested that the Serb IADS may have been unable to deconflict its SAMs and
fighters operating in the same airspace because of identification and discrimination
problems.
19 Barton Gellman, “Key Sites Pounded for 2nd Day,” Washington Post, March 26,
1999. See also John D. Morrocco and Robert Wall, “NATO Vows Air Strikes Will Go the
Distance,” Aviation Week and Space Technology, March 29, 1999, p. 34.
20Hewson, “Operation Allied Force,” p. 17.
21 Telephone conversation with Lieutenant General Michael Short, USAF (Ret.),
August 16, 2001.
Third-night attacks included targets in Mali Mokri Lug, Ayala, Voz-

dovac, and, for the first time, nearer to the immediate outskirts of
Belgrade. That night, 40 percent of the targets attacked were in
Kosovo, as opposed to only 20 percent the first two nights.22 These
attacks, however, just like the ones that took place the preceding two
nights, caused no serious inconvenience for the Serbs. On the con-
trary, the gradually mounting intensity of the air war merely allowed
the Serbs to adjust to a new level of pain, while pressing ahead with
what they had planned all along: to redouble their effort to run as
many ethnic Albanian civilians as possible out of Kosovo and thus be
able to take an unobstructed shot at the KLA once and for all.
This escalated ethnic cleansing should not have come as a complete

surprise to NATO. Weeks earlier, the director of the Central Intelli-
gence Agency (CIA), George Tenet, had predicted that VJ and MUP
forces might respond to a NATO bombing campaign with precisely
such a strategy. Similarly, U.S. military leaders had argued behind
closed doors that air power alone would not suffice to force Milosevic
to back away from such a move.23 The CIA had reportedly learned as
early as fall 1998 that Belgrade was planning a move with tanks and
artillery, called Operation Horseshoe, to drive ethnic Albanians over
the southern and western borders of Kosovo as soon as the snow
melted in the spring. The KLA would thus be stripped of a sur-
rounding civilian population and exposed to direct attack.24 The
Serb incentive for such a move was not difficult to fathom, consider-
ing that the heavily radicalized KLA, which represented the aspira-
tions of most Kosovar Albanians, was (and remains) committed to
the establishment of an independent Kosovo—and a Greater Albania
over the longer term. 25
In any event, Milosevic’s unleashing of large-scale atrocities in

Kosovo and his truculent defiance of NATO denied the alliance the
quick settlement it had counted on and left both NATO and the
______________
23John F. Harris, “Clinton Saw No Alternative to Airstrikes,” Washington Post, April 1,
1999.
24Johanna McGeary, “The Road to Hell,” Time, April 12, 1999, p. 42.
25For an informed treatment of the KLA and its origins, goals, and prospects by The
New York Times’ Balkans bureau chief from 1995 to 1998, see Chris Hedges, “Kosovo’s
Next Masters?” Foreign Affairs, May/June 1999, pp. 24–42.
Clinton administration with no alternative but to continue pressing

the air attacks until NATO unambiguously prevailed. Because
NATO’s leaders on both sides of the Atlantic had banked on a quick
win, no preparations had been undertaken to anticipate what the
consequences might be should Milosevic raise the stakes by acceler-
ating his ethnic cleansing plans. Lest there be any doubt on that
score, General Naumann admitted a month into the bombing that
from the air war’s very start, “there was the hope in the political
camp that this could be over very quickly” and that as a result, no
one at any level had prepared for Milosevic’s ethnic cleansing
push.26 As a result, what had begun as a coercive NATO ploy aimed
at producing Milosevic’s quick compliance quickly devolved into an
open-ended test of wills between the world’s most powerful military
alliance and the wily and resilient Yugoslav dictator.
THE AIR WAR BOGS DOWN

Once it became clear by the fourth day that the air offensive was not
having its hoped-for effect on Milosevic, Clark received authorization
from the NAC to proceed to Phase II, which entailed ramped-up at-
tacks against a broader spectrum of fixed targets in Serbia and
against dispersed and hidden VJ forces in Kosovo. Attacks during the
preceding three nights had focused mainly on IADS targets. Phase II
strikes shifted the emphasis from SEAD to interdiction, with predom-
inant stress on cutting off VJ and MUP lines of communication and
attacking their choke points, storage and marshaling areas, and any
tank concentrations that could be found. Immediately before Phase
II began, NATO ambassadors had argued for more than eight hours,
well past midnight, over whether to expand the target list. General
Naumann insisted at that session that it was time to start “attacking
both ends of the snake by hitting the head and cutting off the tail.”27
His use of that bellicose-sounding metaphor reportedly infuriated
______________
26 Carla Anne Robbins, Thomas E. Ricks, and Neil King, Jr., “Milosevic’s Resolve
Spawned Unity, Wider Bombing List in NATO Alliance,” Wall Street Journal, April 27,
1999. Unlike nearly all other NATO principals, General Naumann cautioned even be-
fore the air war began that although the intent was to be quick, Operation Allied Force
could well turn out to be “long and protracted.” Paul Richter and John-Thor
Dahlburg, “NATO Broadens Its Battle Strategy,” Los Angeles Times, March 24, 1999.
27William Drozdiak, “NATO Leaders Struggle to Find a Winning Strategy,” Washington
Post, April 1, 1999.
the Greek and Italian representatives, who had been calling for an
Easter bombing pause in the hope that it might lead to negotia-
tions.28
NATO went into this second phase earlier than anticipated because
of escalating Serb atrocities on the ground. Up to that point, the air
attacks had had no sought-after effect on Serb behavior whatsoever.
On the contrary, Serbia’s offensive against the Kosovar Albanians
intensified, with Serb troops burning villages, arresting dissidents,
and executing supposed KLA supporters. The Serbs continued un-
opposed in their countercampaign of ethnic cleansing, ultimately
forcing most of the 1.8 million ethnic Albanians in Kosovo from their
homes.29
By the start of the second week, Clinton administration officials

acknowledged that Operation Allied Force had failed to meet its
declared goal of halting Serbian violence against the ethnic
Albanians.30 Echoing that judgment, Clark added that NATO was
confronting “an intelligent and capable adversary who is attempting
to offset all our strategies.”31 It was becoming increasingly clear that
at least one element of Milosevic’s strategy entailed playing for time.
Yet although the humanitarian crime of ethnic cleansing gave the
Serbs an immediate tactical advantage, it also came at the long-term
cost of virtually forcing NATO to stay the course. The bombing effort
thus evolved into a race between those Serb forces trying to drive the
ethnic Albanians out of Kosovo and NATO forces trying to hinder
that effort—or, failing that, to punish Milosevic badly enough to
make him quit.
______________
28The latter rumblings prompted concern in U.S. and NATO military circles that once
any such pause might be agreed to, it would be that much more difficult to resume the
bombing after the pause had expired. In the end, no pause in the bombing occurred
at any time during Allied Force, other than those occasioned by bad weather.
29 After careful examination, the provision of airlift relief missions for the Kosovar
refugees was ruled out by U.S. and NATO planners because they were deemed exces-
sively dangerous in the face of threats from enemy ground fire and because of concern
that any delivered supplies would end up in the wrong hands.
30Bradley Graham and William Drozdiak, “Allied Action Fails to Stop Serb Brutality,”
Washington Post, March 31, 1999.
31Craig R. Whitney, “On 7th Day, Serb Resilience Gives NATO Leaders Pause,” New
York Times, March 31, 1999.
Five B-1Bs were added to the U.S. Air Force’s bomber contingent at
the start of the second week. In preparing them for combat, what
normally would have taken months of effort to program the aircraft’s
mission computers was compressed into fewer than 100 hours dur-
ing a single week as Air Force officers and contractors updated the
computers with the latest intelligence on enemy radar and SAM
threats. One aircraft with the latest updated software installed, the
Block D upgrade of the Defensive System Upgrade Program, passed a
critical flight test at the 53rd Wing at Eglin AFB, Florida, and two B-1s
were committed to action over Yugoslavia two days later. These air-
craft, which, alongside the B-52s, operated out of RAF Fairford in
England, employed the Raytheon ALE-50 towed decoy to good effect
for the first time in combat.32 They were still test-configured aircraft
flown by test crews. The B-1s, all test-configured with Block D up-
grades, typically flew two-ship missions against military area targets,
such as barracks and marshaling yards.
While the Serb pillaging of Kosovo was unfolding on the ground,

NATO air attacks continued to be hampered by bad weather, enemy
dispersal tactics, and air defenses that were proving to be far more
robust than expected. In the absence of a credible NATO ground
threat, which the United States and NATO had ruled out from the
start, VJ forces were able to survive the air attacks simply by spread-
ing out and concealing their tanks and other vehicles. More than half
of the nightly strike sorties returned without any weapons expended
owing to adverse weather in the target area. Only four days out of the
first nine featured weather offering visibility conditions suitable for
employing laser-guided bombs (LGBs). By the end of Day 9, only 15
percent of the 2,700 sorties flown had actually been bombing mis-
sions. In all, it took NATO 12 days to complete the same number of
______________
32Of 10 known Serb SAMs that reportedly guided on B-1s during the course of the air
war, all were believed to have been successfully diverted to the decoys. See David
Hughes, “A Pilot’s Best Friend,” Aviation Week and Space Technology, May 31, 1999, p.
25. The commander of USAFE, General John Jumper, later explained the Serb IADS
tactic employed: Radars in Montenegro would acquire and track the B-1s as they flew
in from over the Adriatic Sea, arced around Macedonia, and proceeded north into
Kosovo. Those acquisition radars would then hand off their targets to SA-6s, whose
radars came up in full target-track mode and fired the missiles, which headed straight
for the ALE-50 and took it out, just as the system was designed to work. “Jumper on
Air Power,” Air Force Magazine, July 2000, p. 43.
strike sorties that had been conducted during the first 12 hours of
Desert Storm.
To all intents and purposes, the difference between Phase II and

Phase I was indistinguishable as far as the intensity of NATO’s air at-
tacks was concerned. The commencement of Phase II was charac-
terized as more of an evolution than a sharp change of direction. On
that point, NATO’s spokesman at the time, RAF Air Commodore
David Wilby, said that the operation was “just beginning to transi-
tion” from IADS targets to fielded VJ and MUP forces.33 By the start
of the second week, merely 1,700 sorties had been flown, only 425 of
which consisted of strike sorties against a scant 100 approved tar-
gets.34 Up to that point, air operations had averaged only 50 strike
sorties a night, in sharp contrast to Desert Storm, in which the daily
attack sortie rate was closer to 1000. The operational goal of Allied
Force was still officially described as merely seeking to “degrade”
Serbia’s military capability. In one of the first hints of growing con-
cern that the air effort was not going well, a senior U.S. general spoke
of at least “several weeks” of needed attacks to beat down VJ and
MUP forces to the breaking point.35 Similarly, by the start of the
second week, an administration official declared that the goal of the
bombing was to “break the will” of the Belgrade leadership, implying
an open-ended air employment strategy.36 Earlier, administration
spokesmen had indicated that they believed that just a few days of
bombing would do the trick.
NATO soon discovered that it was dealing with a cunning opponent

who was quite accomplished at hiding. As a result, it conceded that
it was being forced to “starve rather than shoot them out.”37 Even
with clear skies at the beginning of the third week, NATO pilots were
having little success at interdicting those VJ and MUP troops and
______________
33Bradley Graham, “Bombing Spreads,” Washington Post, March 29, 1999.
34The USAF flew 84 percent of those sorties, the NATO allies 10 percent, and the U.S.
Navy 6 percent.
35Graham, “Bombing Spreads.”
36Craig R. Whitney, “NATO Had Signs Its Strategy Would Fail Kosovars,” New York
Times, April 1, 1999.
37William Drozdiak and Bradley Graham, “NATO Frustration Grows as Mission Falls
Short,” Washington Post, April 8, 1999.
paramilitary thugs in Kosovo who were carrying out the executions,

village burnings, and forced emigration of Kosovar Albanians, to say
nothing of finding and attacking their tanks and artillery. Since at-
tacking dispersed VJ troops directly was proving to be too difficult,
attacks against fielded forces concentrated instead on second-order
effects by going after bases, supplies, and petroleum, oil, and lubri-
cants (POL).
On Day 6, Clark sought NATO approval to increase the pressure on

Milosevic by attacking the defense and interior ministry headquar-
ters in Belgrade. That request was disapproved by NATO’s political
leaders, on the declared ground that such strikes were still
“premature.”38 The list of approved targets increased by about 20
percent at the end of the first week. Yet Clark still did not receive the
full authority from NATO that he had sought. NATO Secretary
General Javier Solana, in particular, expressed misgivings about a
larger target set, saying that he was not persuaded that the time had
come yet to intensify the operation so dramatically. As a result, Clark
was forced to improvise changes to an original plan that had called
for slow-motion escalation, punctuated by pauses, disturbingly
comparable to the flawed strategy employed during Operation
Rolling Thunder over North Vietnam a generation earlier.
Phase III, which entailed escalated attacks against military leader-

ship, command and control centers, weapons depots, fuel supplies,
and other targets in and around Belgrade, commenced de facto on
Day 9 with strikes against infrastructure targets in Serbia. These in-
cluded the Petrovaradin bridge on the Danube at Novi Sad; a bridge
on the Magura-Belacevac railway; the main water supply to Novi Sad;
and targets near Pec, Zatric, Decane, Dragodan, Vranjevac, Bajin
Basta, and the Pristina airport.39 No targets in or near Belgrade,
however, were attacked. At this point, Allied Force was still generat-
ing no more than 50 ground-attack sorties a day.40 There was
mounting unease over the fact that attacks against empty barracks
______________
38 Thomas W. Lippman and Dana Priest, “NATO Builds Firepower for 24-Hour At-
tacks,” Washington Post, March 30, 1999.
39The NAC did not formally approve strikes on Phase III targets per se, although it did
assent to target classes within Phase III.
and other military facilities were having no effect on Serb behavior

now that VJ and MUP forces were well dispersed. It soon became
evident that Milosevic had hunkered down in a calculated state of
siege. Evidently sensing that he had accomplished many of his goals
on the ground and believing that he could now succeed in dividing
NATO, he declared a unilateral cease-fire on April 6. The United
States and NATO, however, rejected that transparent ploy and
pressed ahead with their attacks.
U.S. naval aviation, unavailable for the initial phase of Operation Al-
lied Force, joined the fray when the aircraft carrier USS Theodore
Roosevelt arrived on station in the Ionian Sea south of Italy two weeks
afterward, on April 6. The air wing assigned to the Theodore Roo-
sevelt flew complete and self-sustaining strike packages, including
F-14Ds and F/A-18s for surface-attack operations, EA-6Bs for the
suppression of enemy air defenses, F-14s in the role of airborne
forward air controllers, and E-2Cs performing as ABCCC platforms.
These packages typically flew missions only against dispersed and
hidden enemy forces in Kosovo, although on one occasion, on April
15, they struck a hardened aircraft bunker at the Serbian air base at
Podgorica in Montenegro in the first of several allied efforts to
neutralize a suspected air threat against the U.S. Army’s Task Force
Hawk deployed in Albania (see below).41 The E-2C, normally
operated as an airborne early warning (AEW) platform to screen the
carrier battle group from enemy air threats, was used in Allied Force
to provide an interface between the CAOC and naval air assets
______________
41 Conversation with Vice Admiral Daniel J. Murphy, USN, commander, 6th Fleet,
aboard USS LaSalle, Gaeta, Italy, June 8, 2000. See also Vice Admiral Daniel J. Murphy,
USN, “The Navy in the Balkans,” Air Force Magazine, December 1999, p. 49. According
to a later account by General Jumper, the strike against the Podgorica airfield was the
most concentrated effort placed on any target throughout the entire course of Allied
Force. To satisfy SACEUR’s objective, General Short needed to neutralize the airfield’s
sortie generation capacity completely. At the time the target was selected, only 50 per-
cent of the aim points required to meet that objective had been identified. It took 48
hours to accomplish the additional target analysis and to free up additional required
NATO assets to carry out this strike. Since the Theodore Roosevelt had just arrived in
the theater, it had not been tasked in the April 15 Air Tasking Order and accordingly
had assets that were immediately available. As a result, F-14 and F/A-18 aircraft struck
the hardened aircraft bunker (the highest-value critical element) and used CAOC
(Combined Air Operations Center) assets to assist in targeting and weaponeering.
Other NATO assets struck the remaining critical elements 48 hours later and met
SACEUR’s objectives. Conversation with General John P. Jumper, USAF, Hq Air
Combat Command, Langley AFB, Virginia, May 15, 2001.
operating in the theater, including both strikers and intelligence

collectors. 42
It was hard during the first few weeks for outside observers to assess
and validate the Pentagon’s and NATO’s claims of making progress
because U.S. and NATO officials had so deliberately refrained from
disclosing any significant details about the operation. Instead, ad-
ministration and NATO sources limited themselves to vague general-
izations about the air war’s effects, using such hedged terms as
“degrading,” “disrupting,” and “debilitating” rather than the more
unambiguous “destroying.” On this studiously close-mouthed pol-
icy, the Defense Department’s spokesman, Kenneth Bacon, declared
that a precedent was being intentionally set, since both Secretary
Cohen and General Shelton had seen a need to “change the culture
of the Pentagon and make people more alert to the dangers that can
flow from being too generous—or you could say profligate or lax—
with operational details.”43
In one of the first tentative strikes against enemy infrastructure, the

main telecommunications building in Pristina, the capital city of
Kosovo, was taken out by two GBU-20 LGBs dropped by an F-15E on
April 6. Yet the air effort as a whole remained but a faint shadow of
Operation Desert Storm, with only 28 targets throughout all of Yu-
goslavia attacked out of 439 sorties in a 24-hour period during the
operation’s third week.44 As for the hoped-for “strategic” portion of
the air war against the Serb heartland, Clark was still being refused
permission by NATO’s political leaders to attack the state-controlled
television network throughout Yugoslavia. On April 12, attacks were
______________
42See Commander Wayne D. Sharer, USN, “The Navy’s War over Kosovo,” Proceed-
ings, U.S. Naval Institute, October 1999, pp. 26–29; and Robert Wall, “E-2Cs Become
Battle Managers with Reduced AEW Role,” Aviation Week and Space Technology, May
10, 1999, p. 38.
43Jason DeParle, “Allies’ Progress Remains Unclear as Few Details Are Made Public,”
New York Times, April 5, 1999.
44In fairness to that effort, however, and given the many constraints that affected it—
in contrast to the far fewer constraints that affected Desert Storm—weather, mainly an
irritant during the Gulf War, was a significant factor during Operation Allied Force.
Bad weather, combined with the higher population density of Serbia, the concern for
collateral damage, and the increased surface-to-air threat, could easily have con-
tributed to a lower relative intensity of strike operations. I thank Major Richard
Leatherman, Hq Air Force Doctrine Center, for having called this possibility to my at-
tention.
conducted against an oil refinery at Pancevo and other infrastructure

targets, with the Pentagon announcing that all of Yugoslavia’s oil re-
fineries had been destroyed but that some stored fuel remained
available. Also on April 12, the 20th day of the air attacks, NATO
missions into the newly designated Kosovo Engagement Zone (KEZ)
commenced with attempted attacks against VJ and MUP tanks, ar-
tillery, wheeled vehicles, and other assets fielded in Kosovo, in re-
sponse to Belgrade’s escalated ethnic cleansing of the embattled
province.
By the third week, NATO’s strategic goals had shifted from seeking to
erode Milosevic’s ability to force an exodus of Kosovar Albanian
civilians to enforcing a withdrawal of Serb forces from Kosovo and a
return of the refugees home. That shift in strategy was forced by
Milosevic’s early seizure of the initiative and his achievement of a
near–fait accompli on the ethnic cleansing front. Up to that point,
President Clinton had merely insisted that the operation’s goal was
to ensure that Milosevic’s military capability would be “seriously
diminished.”45
As Operation Allied Force continued to bog down entering its third

week, the influential London Economist pointedly observed that it
was not “just NATO whose credibility is at risk. At home, the Defense
Department’s post-Gulf-war prestige is also in the balance, along
with the doctrines of high-tech dominance that the Gulf war encour-
aged people to believe. America’s faith in air power, formed by the
precision bombing of Iraqi targets, has already been tested by Mr.
Hussein’s durability. If the current bombing fails to unseat Mr. Milo-
sevic, the air power doctrine could collapse.”46 Numerous factors
accounted for why the operation’s early performance had proven so
disappointing. They included adverse weather, difficult terrain, a
wily and determined opponent, poor strategy choices by the Clinton
administration and NATO’s political leaders, and, perhaps most of
all, the burdens of having to coordinate an air operation with 18 of-
ten highly independent-minded allies.
______________
45John M. Broder, “Clinton Says Milosevic Hurts Claim to Kosovo,” New York Times,
March 31, 1999.
46“Hope for the Best, and a Spot of Golf,” The Economist, April 3, 1999, p. 9.
By the end of the third week, in large measure out of frustration over
the operation’s continued inability to get at the dug-in and elusive VJ
positions in Kosovo, Clark requested a deployment of 300 more air-
craft to support the effort. That request, which would increase the
total number of committed U.S. and allied aircraft to nearly 1,000,
entailed more than twice the number of allied aircraft (430) on hand
when the operation began on March 24—and almost half of what the
allied coalition had had available for Desert Storm. For the United
States, it represented a 60 percent increase over the 500 U.S. aircraft
already deployed (see Figure 3.1 for the ultimate proportions of U.S.
and allied aircraft provided to support Allied Force). Among other
things, it prompted understandable concern about where to base
RAND MR1365-3.1
USN/USMC
5%
Allied 41% 54% USAF
SOURCE: AWOS Fact Sheet.
Figure 3.1—U.S. and Allied Aircraft Contributions

them, with NATO looking to France, Germany, Hungary, Turkey, and

the Czech Republic for possible options.47
The call for 300 additional aircraft followed on the heels of an earlier
request by Clark for 82 more aircraft, which had been promptly ap-
proved by the Pentagon. This time, Pentagon officials expressed
surprise at the size of Clark’s request and openly questioned whether
it would be approved in its entirety. 48 The principal concern was
that it would draw precious assets, notably such low-density/high-
demand aircraft as the E-3 airborne warning and control system
(AWACS) and EA-6B Prowler, away from Iraq and Korea. The service
chiefs reportedly complained that Clark’s requested quantities repre-
sented a clear case of overkill and that USEUCOM was not making
the most of the forces already at its disposal.
In addition, Clark asked for the USS Enterprise and its 70-aircraft air
wing, which would necessitate extending the carrier’s cruise length
and thereby breaking a firm Navy rule of not keeping aircrews and
sailors at sea for any longer than six months at a single stretch. The
request was opposed by the chief of naval operations, Admiral Jay
Johnson, and Secretary of the Navy Richard Danzig. In the end, the
Enterprise was made available by the Navy for diversion to the
Adriatic as requested, but its air wing was never tasked by the CAOC,
and it never participated in Allied Force. Once the additional aircraft
were approved, NATO asked Hungary to make bases available and
Turkey to help absorb those aircraft. Figure 3.2 shows how the in-
theater buildup of aircraft ultimately played itself out.
As a part of his requested force increment, Clark also asked for a de-
ployment of Army AH-64 Apache attack helicopters. Although the
other aircraft were eventually approved by Secretary Cohen and the
Joint Chiefs of Staff (JCS), this particular request was initially
______________
47 Steven Lee Myers, “Pentagon Said to Be Adding 300 Planes to Fight Serbs,” New
York Times, April 13, 1999.
48One former senior U.S. officer commented that Clark had presented “a wish list that
would choke a horse.” Elaine M. Grossman, “Clark’s Firepower Request for Kosovo
Prompts Anxiety Among Chiefs,” Inside the Pentagon, April 15, 1999.
RAND MR1365-3.2
1,200
1,000
800
Number of aircraft
USAF
600
400
USA, USN,
USMC
200
Allies
0
February 23
January 1 Rambouillet March 24 April 23 June 10
talks end Allied NATO summit Cease-fire
Force
begins
SOURCE: Hq USAFE/SA.
Figure 3.2—In-Theater Aircraft Buildup
disapproved, on the avowed premise that since attack helicopters are

typically associated with land combat, the introduction of Apaches
might be misperceived by some allies as a precursor to a ground
operation. A more serious concern almost surely was that the
Apaches might not survive were they to be committed to combat in
the still-lethal Serb SAM and AAA environment.
In the end, despite Army and JCS reluctance, Clark prevailed in his
request for the Apaches and announced that 24 would be deployed
to Albania from their home base at Illesheim, Germany. Pentagon
spokesmen went out of their way to stress that the Apaches were in-
tended solely as an extension of the air effort and not as an implied
prelude to future ground operations.49 To support the major aircraft

buildup requested by Clark, the Pentagon asked President Clinton to
authorize a call-up of 33,000 U.S. reservists and National Guard
members—in the largest single reserve-force activation since Desert
Shield in 1990–1991, when 265,000 Guard and reserve personnel had
been mobilized for possible combat or combat-support duty in the
Gulf. As the bombing entered its 27th day, Clinton asked Congress
for $5.458 billion in emergency funding to continue financing the air
effort, with $3.6 billion to cover air operations from March 24 to the
end of FY99, $698 million for additional cruise missiles and preci-
sion-guided munitions (PGMs), and $335 million for refugee relief.
Meanwhile, Operation Allied Force remained as stalled as ever, with

no sign of tangible progress. Clark had wanted to go after command
bunkers and other vital targets in Serbia from the very start of the op-
eration, but it took more than a month for NATO’s political leaders to
approve an attack on Milosevic’s villa at Dobanovci. The Dutch gov-
ernment steadfastly refused to grant approval to bomb his presiden-
tial palace because the latter was known to contain a painting by
Rembrandt.50 NATO’s ambassadors would not even approve strikes
against occupied VJ barracks out of expressed concern over causing
too many casualties among helpless enemy conscripts.51
By the end of the first month, as many as 80 percent of the strikes

conducted had been revisits to fixed targets that had been attacked at
least once previously. This was due in part to rapid enemy regenera-
tion and reconstitution efforts, but mainly to the limited number of
targets that had been approved by NATO’s civilian leaders, the often
maddeningly slow target generation and approval process, and
______________
49Bradley Graham and Dana Priest, “Allies to Begin Flying Refugees Abroad,” Wash-
ington Post, April 5, 1999. In his subsequent memoirs, Clark frankly excoriated what
he called “the reluctant Army mind-set in Washington” on this issue and dismissed
other critics of his requested AH-64 commitment as “the voices of conventional air
power” coming from “commanders who had no experience with the Apaches.” Gen-
eral Wesley K. Clark, Waging Modern War: Bosnia, Kosovo, and the Future of Combat,
New York, Public Affairs, 2001, pp. 230, 278.
50To which General Naumann countered in resigned exasperation: “It isn’t a good
Rembrandt.” Robbins, Ricks, and King, Jr., “Milosevic’s Resolve Spawned More Unity
in Alliance and a Wider Target List.”
51Michael Ignatieff, “The Virtual Commander: How NATO Invented a New Kind of
War,” The New Yorker, August 2, 1999, p. 32.
SACEUR’s desire to keep the bombs falling on Serbia notwithstand-

ing those constraints. The Serbs repeatedly demonstrated an ability
to perform workarounds and rebuild their damaged communica-
tions facilities, with some IADS installations being brought back on
line less than 24 hours of having been attacked. 52 Furthermore,
among all the Air Tasking Order (ATO) releases of combat aircraft
against preassigned targets, some failed to get airborne because of
weather cancellations or maintenance-related aborts, and others ei-
ther returned home without having expended their ordnance be-
cause of rules-of-engagement constraints or having failed to hit their
assigned aim points when they did succeed in dropping munitions.
These considerations also figured prominently in the low effective-
ness of the overall effort.
As the air war entered its fifth week, Clark admitted that Milosevic
was still pouring reinforcements into Kosovo continuously and that
“if you actually added up what’s there on a given day, you might find
that he’s strengthened his forces in there.” 53 Much as during some
periods of Desert Storm, adverse weather at the five-week point had
forced a cancellation or failure of more than half of all scheduled
bombing sorties on 20 of the first 35 days of air attacks. Seemingly
resigned to a waiting game as the air war appeared stalled after more
than a month of continual bombing, a senior NATO diplomat con-
fessed that it now felt as though Operation Allied Force “had been
put on autopilot. Now we are basically waiting for something to
crack in Belgrade.”54 In light of the stalled offensive, some saw the
air war now threatening to stretch into summer 1999, if not longer.
______________
52Bradley Graham and John Lancaster, “Most NATO Bombing Raids Target Previously
Hit Sites,” Washington Post, April 21, 1999. In fairness to NATO planners, some of
those reattacks were valid, because a few especially large area targets entailed nu-
merous individual aim points, some of which were missed in the initial attacks. The
vast majority, however, merely entailed what many frustrated NATO crewmembers re-
ferred to as “bouncing rubble,” having no practical effect and presenting considerable
added risk to their own survivability. It was not uncommon for aircrews to complain
vocally about having their “warm bodies sent out all over again to turn bricks into
powder.”
53Craig R. Whitney, “NATO Chief Admits Bombs Fail to Stem Serb Operations,” New
54Neil King, Jr., “War Against Yugoslavia Lapses into Routine, but Clock Is Ticking,”
Wall Street Journal, May 6, 1999.
NATO FINALLY ESCALATES

In what proved in hindsight to be a watershed development for Op-
eration Allied Force, the NATO summit that convened in Washington
on April 23–25 to commemorate the alliance’s 50th anniversary was
pivotal in solidifying NATO’s collective determination not to lose. As
President Clinton’s national security adviser, Samuel Berger, later
attested, NATO’s leaders unanimously agreed at the summit that “we
will not lose. We will not lose. Whatever it takes, we will not lose.”55
Part of the mounting pressure on U.S. and NATO leaders to show
greater resolution emanated from a public mood on both sides of the
Atlantic that was growing increasingly sensitive to, and emboldened
by, the horrific privations inflicted on helpless Kosovar Albanians by
their Serb oppressors, shown daily on worldwide television—a public
reaction, one might add, that calls into serious question the oft-
heard assertion that Milosevic “won” the media campaign. The ugly
spectacle of the ethnic cleansing push finally drove the allied leaders
to turn the corner at the Washington summit, after which, as General
Jumper later observed, “we really had the level of consensus we
should have had to start this thing off. . . . After the Washington
summit, there was no way that NATO was going to let itself fail.”56
That consensus, along with the refugee crisis, occasioned an in-

creased NATO willingness to attack major infrastructure targets.
Eventually, thanks to this heightened inclination to ramp up the
pressure, NATO’s Master Target File grew from only 169 targets on
the eve of the air effort to more than 976 by its end in early June.57
Once the call for a substantially expanded target list had prevailed,
the new goal became punishing Belgrade’s political and military
elites, weakening Milosevic’s domestic power base, and demonstrat-
______________
55Doyle McManus, “Clinton’s Massive Ground Invasion That Almost Was,” Los Ange-
les Times, June 9, 2000.
56General John Jumper, USAF, “Oral Histories Accomplished in Conjunction with
Operation Allied Force/Noble Anvil.”
57Dana Priest, “Target Selection Was Long Process,” Washington Post, September 20,
1999. One must take care, however, not to confuse Master Target File growth with
approved target growth. Although target nominations increased dramatically as the
air war entered full swing, getting those targets individually approved remained a
challenge throughout the air war to the very end.
ing by force of example that he and his fellow perpetrators of the

abuses in Kosovo would find no sanctuary.
Even before the Washington summit, NATO’s targeting efforts had

already begun to focus gradually not just on dispersed and hidden
enemy forces in Kosovo, but also on what NATO officials had come
to characterize as the four pillars of Milosevic’s power—the political
machine, the media, the security forces, and the economic system.
New targets added to the approved list included national oil refiner-
ies, petroleum depots, road and rail bridges over the Danube, railway
lines, military communications sites, and factories capable of pro-
ducing weapons and spare parts.58 The first attacks against state
radio and television stations in Belgrade took place on April 21, with
three cruise missiles temporarily shutting down three channels run
by Milosevic’s wife, Mira Markovic, destroying the 12th through the
17th floors of the building, and killing several journalists and techni-
cians, after NATO had issued a warning to employees to vacate the
buildings. (Transmissions resumed 11 hours later, occasioning a
reattack.) With that escalation, NATO finally brought the air war to
Yugoslavia’s political and media elite after weeks of hesitation, indi-
cating that it was now emboldened enough to go directly after the
business interests of Milosevic’s family and friends. In the same at-
tack, U.S. cruise missiles took out the offices of the political parties of
Milosevic and his wife. Also on April 21, the Zezel bridge, the last
remaining bridge over the Danube at Novi Sad, was dropped.
On April 28, a large, coordinated attack was launched against the
Serb military airfield at Podgorica, with 30 munitions employed
against such targets as hardened shelters, POL facilities, radar sites,
and aircraft and helicopters parked in revetments. During that at-
tack, a 4,700-lb GBU-28 “bunker-buster” was dropped for the first
time in Allied Force by an F-15E on an underground aircraft and
equipment storage hangar at the Pristina airfield. (By that point in
the air war, F-15Es had begun flying seven-and-a-half-hour missions
into Serbia directly from RAF Lakenheath in England.)59 Having
______________
58 Eric Schmitt and Steven Lee Myers, “NATO Said to Focus Raids on Serb Elite’s
Property,” New York Times, April 19, 1999.
59In the end, however, only some 10 percent of the 48th Fighter Wing’s F-15E combat
missions were flown out of Lakenheath. The remainder were flown out of the wing’s
been repeatedly attacked before with less destructive munitions, that

buried hangar and the remaining aircraft, munitions, and supplies
kept in it were thought to have been taken out once and for all by this
weapon, an assessment which later proved false.60 Shortly there-
after, an attack was conducted against the national command center
in Belgrade, a multistory facility buried more than 100 feet under-
ground and known to have been one of Milosevic’s occasional re-
treats.61 Equipped with communications, medical facilities, living
spaces, and enough food to last more than a month, it was designed
to accommodate the entire Yugoslav general staff, top defense offi-
cials, and other civilian authorities.62
Despite these ramped-up attacks, however, the French leadership

remained critical of many proposed strike options. In particular,
President Jacques Chirac opposed any attacks against Belgrade’s
electrical power grid with high-explosive bombs that would physi-
cally render it inoperative for any length of time. In an effort to get
around Chirac’s resistance, U.S. planners worked behind the scenes
with French officers in search of more palatable alternatives. As re-
ported in a later U.S. press account, they finally came up with the
idea of using the CBU-104(V)2/B cluster munition, formerly referred
to by some U.S. Air Force officers as the CBU-94, which could shut
down Belgrade’s power source for at least a few hours by depositing
carbon-graphite threads on the electrical grid, an option to which
Chirac finally consented.63
Thanks to that modest breakthrough, in possibly the most conse-

quential attack of the air war up to that point, USAF F-117s report-
edly dropped CBU-104s on five transformer yards of Yugoslavia’s
electrical power grid—at Obrenovac, Nis, Bajina Basta, Drmno, and
_____________________________________________________________
forward operating location at Aviano. Conversation with USAF F-15E aircrews, 492nd
Fighter Squadron, RAF Lakenheath, England, April 28, 2001.
60Robert Hewson, “Allied Force, Part II: Overwhelming Air Power,” World Air Power
Journal, Winter 1999/2000, p. 99.
61Paul Richter, “Bunker-Busters Aim at Heart of Leadership,” Los Angeles Times, May
5, 1999.
62Ibid. The Pentagon’s formal report to Congress later indicated that “some” hard-
ened underground command bunkers had been destroyed.
63 Dana Priest, “France Acted as Group Skeptic,” Washington Post, September 20,
1999, and David A. Fulghum, “Russians Analyze U.S. Blackout Bomb,” Aviation Week
and Space Technology, February 14, 2000, p. 59.
Novi Sad—during the early morning hours of May 3, temporarily

cutting off electricity to 70 percent of the country. These munitions
were similar to weapons delivered by TLAMs against the Baghdad
electrical power network during the opening hours of Operation
Desert Storm. The effects were achieved by means of scattered reels
of treated wire which unwound in the air after being released as
BLU-114/B submunitions, draping enemy high-voltage power lines
like tinsel and causing them to short out.64 The announced intent of
that escalated attack was to shut down the installations that provided
electrical power to the VJ’s 3rd Army in Kosovo to disrupt military
communications and confuse Serb air defenses.65 Very likely an
unspoken intent was also to tighten the air operation’s squeeze on
the Serbian political leadership and rank and file. 66
Whatever the case, the attack moved NATO over a new threshold and
brought the war, for the first time, directly to the Serbian people. By
the end of the seventh week, there began to be reports of Yugoslav
officials openly admitting that the country was on the verge of
widespread hardship because of the air war’s mounting damage to
the nation’s economy, which had already been weakened by almost
four years of international sanctions imposed for Serbia’s earlier role
in the war in Bosnia.67 The destruction of one factory in Krujevac
that produced automobiles, trucks, and munitions resulted in 15,000
people being put out of work, plus 40,000 more who were employed
by the factory’s various subcontractors. Attacks against other facto-
ries had similar effects on the Yugoslav economy. By the time Allied
Force had reached its halfway point, the bombing of infrastructure
targets had halved Yugoslavia’s economic output and deprived more
______________
64An inertial navigation system (INS)–guided version of the weapon, a variant of the
wind-corrected munitions dispenser, is now said to be entering the U.S. munitions in-
ventory. Fulghum, “Russians Analyze U.S. Blackout Bomb.”
65David A. Fulghum, “Electronic Bombs Darken Belgrade,” Aviation Week and Space
Technology, May 10, 1999, p. 34.
66The results were more symbolic than strategically significant. After the May 3 at-
tack, some 500 workers managed to clear the filaments sufficiently to restart the
equipment within 15 hours. After a similar attack on May 8, the threads were cleared
within 4 hours. William Arkin, “Smart Bombs, Dumb Targeting?” Bulletin of the
Atomic Scientists, May/June 2000, p. 52.
67 Robert Block, “In Belgrade, Hardship Grows Under Sustained Air Assault,” Wall
Street Journal, May 12, 1999.
than 100,000 civilians of jobs. Local economists reported that the

effect was more damaging than that of the successive Nazi and allied
bombing of Yugoslavia during World War II, when the country was
far more rural in its economic makeup. A respected economist at
Belgrade University who coordinated a group of economists from the
International Monetary Fund (IMF) and World Bank, Mladjan
Dinkic, called the results of the bombing an “economic catastrophe,”
adding that while the Serb population would not die of hunger, “our
industrial base will be destroyed and the size of the economy cut in
half.” 68
Only during the last two weeks of Allied Force, however, did NATO
finally strike with real determination against Serbia’s electrical power
generating capability, a target set that had been attacked in Baghdad
from the very first days of Desert Storm. The earlier “soft” attacks at
the beginning of May with graphite filament bombs against the
transformer yards of Yugoslavia’s main power grid had caused a
temporary disruption of the power supply by shorting out trans-
formers and disabling them rather than destroying them. But this
time, in perhaps the single most attention-getting strike of the entire
air war up to that point, the Yugoslav electrical grid was severely
damaged over the course of three consecutive nights starting on May
24. Those attacks, directed against electrical power facilities and re-
lated targets in Belgrade, Novi Sad, and Nis, the three largest cities in
Serbia, shut off the power to 80 percent of Serbia, leaving millions
without electricity or water service. They affected the heart of Yu-
goslavia’s IADS, as well as the computers that ran its banking system
and other important national consumers of electricity.69
As evidence that these infrastructure attacks were making their ef-

fects felt, the early street dancing and carefully orchestrated demon-
strations of studied outrage against NATO in response to its earlier
pinprick attacks became displaced by a manifest weariness on the
part of most residents. Clark continued to stress that the top priority
was to destroy the VJ’s 3rd Army or run it out of Kosovo. He also ac-
______________
68Steven Erlanger, “Economists Find Bombing Cuts Yugoslavia’s Production in Half,”
69Philip Bennett and Steve Coll, “NATO Warplanes Jolt Yugoslav Power Grid,” Wash-
ington Post, May 25, 1999.
knowledged, however, the goal of disrupting the everyday life of Serb

citizens.70 By late May, NATO military commanders had received
authorization to attack Yugoslavia’s civilian telephone and computer
networks in an effort to sever communications between Belgrade
and Kosovo.71 In all of this, a long-discredited premise of classic air
power theory, namely, that the bombing of civilian infrastructure
would eventually prompt a popular reaction, seemed to be showing
some signs of validity. Until that key turning point, Clark later ob-
served, Operation Allied Force had been “the only air campaign in
history in which lovers strolled down riverbanks in the gathering twi-
light and ate at outdoor cafes and watched the fireworks.”72
FACING THE NEED FOR A GROUND OPTION

During the air war’s initial weeks, administration officials continued
to adhere to their initial hope that an air effort alone would eventu-
ally elicit the desired response from Milosevic. Even after Allied
Force was well under way, Secretary General Solana announced that
he was sure that the bombing would be over before the start of the
long-planned Washington summit on April 23 to celebrate NATO’s
50th birthday. 73 Deep doubts that the air attacks alone would suffice
in forcing Milosevic to knuckle under, however, soon prompted a
steady rise in military pressure—notably from some U.S. Air Force
leaders directly involved with the air war—for developing at least a
fallback option for a ground invasion.74
Indeed, even before the operation was a week old, indications had
begun to mount that senior administration officials were starting to
have second thoughts about the advisability of having peremptorily
______________
70Ibid.
71William Drozdiak, “Allies Target Computer, Phone Links,” Washington Post, May 27,
1999.
72Ignatieff, “The Virtual Commander,” p. 35.
73James Gerstenzang and Elizabeth Shogren, “Serb TV Airs Footage of 3 Captured U.S.
Soldiers,” Los Angeles Times, April 1, 1999.
74On that account, Clark later acknowledged that his air commanders were no hap-
pier than he was with the absence of a ground threat, noting that it was “sort of an un-
natural act for airmen to fight a ground war without a ground component.” Ignatieff,
“The Virtual Commander,” p. 33.
ruled out a ground option before launching into Allied Force. The
chairman of the JCS, General Shelton, for example, remarked that
there were no NATO plans “right now” to introduce ground troops
short of a peace settlement in Kosovo.75 In a similarly hedged re-
mark, Secretary Cohen pointed out that the Clinton administration
and NATO had no plans to introduce any ground troops “into a hos-
tile environment,” leaving open the possibility that they might con-
template putting a ground presence into a Kosovo deemed
“nonhostile” before the achievement of a settlement. 76 By the end of
the second week, Secretary of State Albright went further yet toward
hinting at the administration’s growing discomfiture over having
ruled out a ground threat when she allowed that NATO might change
its position and put in ground troops should the bombing succeed in
creating a “permissive environment.”77 Ultimately, the air war’s
continued indecisiveness led President Clinton himself to concede
that he would consider introducing ground troops if he became
persuaded that the bombing would not produce the desired
outcome. In a clear contradiction to his earlier position on the issue,
he asserted that he had “always said that . . . we have not and will not
take any option off the table.” That statement was later described by
a U.S. official as testimony to an ongoing administration effort “to
break out of a rhetorical box that we should never have gotten
into.” 78
By the start of the third week, a consensus had begun to form in

Washington that ground forces might well be needed, if only to sal-
vage NATO’s increasingly shaky credibility that was being steadily
______________
75 Paul Richter, “Use of Ground Troops Not Fully Ruled Out,” Los Angeles Times,
March 29, 1999.
76Rowan Scarborough, “Military Experts See a Need for Ground Troops,” Washington
77Rowan Scarborough, “Momentum for Troops Growing,” Washington Times, April 5,
1999.
78John F. Harris, “Clinton Says He Might Send Ground Troops,” Washington Post, May
19, 1999. In an earlier attempt at revisionism, Secretary of State Albright upbraided an
interviewer by flatly declaring that “we never expected this to be over quickly,” in
complete contradiction to her categorical pronouncement the first night of the air war
11 days earlier that “I think that this is something, the deter and damage, is something
that is achievable within a relatively short time.” John Harris, “Reassuring Rhetoric,
Reality in Conflict,” Washington Post, April 8, 1999.
eroded by the air operation’s lackluster performance.79 That

dawning realization led to two parallel escalation processes: one
highly public—the substantial increase in the number of committed
aircraft, the growing number of approved targets and heightened
percentage of daily shooter sorties, and the hard attacks conducted
against the Yugoslav power grid; and the other largely beneath public
scrutiny—namely, more serious discussion within the U.S. govern-
ment over the need to begin making concrete preparations for a
ground intervention. Deputy Secretary of State Strobe Talbott con-
veyed a “very explicit” warning to Russian envoy Vladimir Cher-
nomyrdin that President Clinton was seriously considering a ground
option, a warning which we can assume Chernomyrdin duly passed
on to Milosevic.
At the same time, a pronounced rift emerged between Clark and his
Pentagon superiors over Clark’s insistence on replacing talk with de-
termined action in connection with preparations for a ground inva-
sion. In his memoirs, Clark later gave candid vent to his frustration
over this rift when he referred to the “divide between those in
Washington who thought they understood war and those [of us] in
Europe who understood Milosevic, the mainsprings of his power,
and the way to fight on this continent.”80 Earlier in April, he had
challenged U.S. and British officers at NATO headquarters to con-
sider “what if” options for a potential ground war. Out of frustration
over the refusal of both Washington and his NATO masters to coun-
tenance any serious consideration of a ground component to Allied
Force, he also asked the Army, shortly after the air war began, to send
him a half-dozen officers from the School of Advanced Military
Studies at Fort Leavenworth, Kansas, to draw up secret plans for a
broad spectrum of ground options, ranging from sending in peace-
keepers to police any settlement that might be achieved single-
handedly by the air war to launching a full-fledged, opposed-entry
land invasion if all else proved wanting. It soon became clear from
______________
79 Dan Balz, “U.S. Consensus Grows to Send in Ground Troops,” Washington Post,
April 6, 1999.
80Clark, Waging Modern War, p. 303. As a testament to the depth of his conviction on
the criticality of getting serious about laying the groundwork for a land invasion, Clark
in mid-May wrote a letter to Secretary General Solana which, he said, “demonstrated
at length how moving into ground-force preparations would exponentially increase
[NATO’s] leverage against Milosevic.” Ibid, pp. 307–308, emphasis added.
that inquiry that only about a dozen roads led into Kosovo from Al-
bania. Like Kosovo’s bridges, they were heavily mined and strongly
defended, with VJ troops well positioned on the high ground of the
most strategically crucial terrain. Accordingly, the study concluded
that the best invasion routes would be from Hungary and Croatia
into the flatter terrain of northern Serbia.
Several administration officials later commented that an invasion

threat from both east and west (namely, from Romania and Croatia
and from Hungary) would have been preferable to one from Albania
alone, where the transportation infrastructure was extremely primi-
tive and where wheeled vehicles would quickly bog down in wet
weather.81 They further characterized the nascent ground threat as
pointed not just at Kosovo but at Serbia proper, since such an opera-
tion would aim directly at Milosevic’s greatest vulnerability and, in so
doing, threaten to take down his regime. Secretary General Solana
later allowed that he had authorized NATO’s military command to
revise and update plans for a possible ground invasion, while at the
same time indicating that the alliance was still far from any decision
to use ground forces and voicing his conviction that the air effort
would ultimately achieve its objectives.82
Throughout this secret ground-options planning, Clark was strongly

resisted by Secretary Cohen and White House security adviser
Berger. But he had the unwavering support of Britain’s prime minis-
ter, Tony Blair, who had unsuccessfully raised the issue of ground
troops with Clinton at the NATO summit in late April. Not long
thereafter, apparently reflecting growing British concern that air at-
tacks alone would prove insufficient to compel Milosevic to quit,
British Foreign Minister Robin Cook took the lead in mid-May in
proposing that allied ground troops be sent into Kosovo, even in the
______________
81Interview by RAND staff, Washington, D.C., June 11, 2000. The UK Ministry of De-
fense’s director of operations in Allied Force, Air Marshal Sir John Day, however, later
commented that there was never much military enthusiasm for a double envelopment
through Hungary. Conversation with Air Marshal Day, RAF Innsworth, United King-
dom, July 26, 2000.
82Thomas W. Lippman and Bradley Graham, “NATO Chief Asks Review of Invasion
Planning,” Washington Post, April 22, 1999.
absence of a peace agreement, once the bombing had reduced VJ

forces to a point where they could mount only scattered resistance.83
After a month of continued inconclusiveness in the air war, one be-

gan to hear talk not only at NATO headquarters but also in Washing-
ton regarding the need for a credible ground threat to evict the ma-
rauding Serb forces from Kosovo. As the end of the second month
approached, NATO appeared more than ever headed toward conced-
ing at least the possibility of a land invasion, even though the Clinton
administration would still not brook even a hint of encouraging pub-
lic debate over the subject. As one possible explanation for the ad-
ministration’s continued reluctance to embrace the growing need for
a ground operation of some sort, polls taken during the air war’s sev-
enth week indicated that war fatigue was setting in, occasioning the
first significant decline in U.S. public support. That support dropped
from 65 percent in late April to 59 percent by mid-May, with opposi-
tion to the air war rising from 30 to 38 percent during the same pe-
riod.84
By late May, with winter weather promising to become a limiting fac-

tor as early as the beginning of October, Clark had begun to stress
that time was now critical with respect to planning for a ground in-
vasion. On one occasion, he expressly warned NATO’s civilian lead-
ers and Washington alike: “Don’t let the decision make itself.”85 In
the British view, September 15 was absolutely the latest date on
which a ground push could start, based on a determination that it
would take a minimum of one month to complete the operation.86
By most accounts, the turning point in facing up to the need for a se-
rious ground option came on May 27, when Cohen met secretly in
Bonn with his four principal NATO counterparts, the British, French,
German, and Italian defense ministers, in a six-and-a-half-hour ses-
______________
83Eric Schmitt and Michael R. Gordon, “British Pressing Partners to Deploy Ground
Troops,” New York Times, May 18, 1999.
84Richard Morin, “Poll Shows Most Americans Want Negotiations on Kosovo,” Wash-
85Carla Anne Robbins and Thomas E. Ricks, “Time Is Running Out If Invasion Is to
Remain Option Before Winter,” Wall Street Journal, May 21, 1999.
86Conversation with Air Marshal Sir John Day, RAF, UK Ministry of Defense director of
operations in Allied Force, RAF Innsworth, United Kingdom, July 26, 2000.
sion convened expressly to consider what it would take to mount a

land invasion and to weigh the merits and risks of such a course of
action. By one informed account, that meeting was pivotal in getting
the allies to come to closure once and for all on the need to begin se-
rious preparations for a land invasion.87 The chief of Britain’s de-
fense forces, General Sir Charles Guthrie, was an especially strong
backer of a ground option, as was the British defense minister,
George Robertson. The RAF also had agreed from the start that a
ground option was needed. As but one indicator that acceptance of
the need for such an option had, by that time, become all but a fait
accompli, British planning had progressed to the point of actual re-
serve call-up and the booking of ferries and civil air transports to de-
liver British troops to the combat zone. The Bonn ministerial meet-
ing thus took the process begun at the NATO summit a step further
toward solidifying the idea that NATO was going to win, come what
may, by extending that notion to include acceptance of a ground in-
vasion should matters come to that.
By the end of May, NATO was generally acknowledged by the media

as “inching ever closer to some kind of ground operation in the
Balkans.”88 Lending further credence to that impression, several
administration officials later acknowledged that Britain was on board
with the United States by that time for a ground invasion if need be.
They further acknowledged that most of the allies’ concerns about
attacking infrastructure targets had been largely put to rest, even
though France remained an obstacle, and that it would have been
easier to obtain NAC approval to go after increasingly sensitive tar-
gets with air attacks once the likelihood of a NATO ground invasion
loomed larger.89
COUNTDOWN TO CAPITULATION
Following an inadvertent attack on a refugee convoy near Djakovica,
Kosovo, on April 14—occasioned in part by a suspected visual
misidentification by the participating USAF F-16 pilots (see Chapter
______________
87Ibid.
88Richard J. Newman, “U.S. Troops Edge Closer to Kosovo,” U.S. News and World Re-
port, June 7, 1999.
89Interview by RAND staff, Washington, D.C., July 11, 2000.
Six)—the altitude floor of 15,000 ft that had been imposed at the start
of the air war was eased somewhat in the southern portion, and
NATO forward air controllers (FACs) flying over Kosovo were cleared
to descend to as low as 5,000 ft if necessary, to ensure positive iden-
tification of ground targets in the KEZ.90 Direct attacks on suspected
VJ positions in Kosovo by B-52s occurred for the first time on May 5
and again the following day. Clark declared afterward that 10 enemy
armor concentrations had been hit and that the Serbs were no longer
able to continue their ethnic cleansing. NATO spokesmen further
reported that enemy troops in the field were running low on fuel and
that VJ and MUP morale had declined.91 A day later, NATO claimed
that it had destroyed 20 percent of the VJ’s artillery and armor de-
ployed in Kosovo. As for infrastructure attacks, only two of the 31
bridges across the Danube in Yugoslavia were said to be still func-
tional by the end of the week. During the second week of May, how-
ever, enemy attack helicopters conducted an attack against the vil-
lage of Kosari along the main supply route for the KLA. They also
served as spotters for VJ artillery against KLA pockets of resistance.92
Those operations indicated that NATO had done an imperfect job of
preventing any and all enemy combat aircraft from flying.
On May 12, roughly 600 Allied Force sorties were launched all told,
including the highest daily number of shooter sorties to date. (See
Figure 3.3 for the overall trend line in U.S. and allied sorties flown
over the 78-day course of Allied Force. Most of the troughs in that
trend line indicate sortie drawdowns or cancellations occasioned by
nonpermissive weather over Serbia.) The multiple waves of succes-
sive large force packages commenced with a sunrise launch of 36 air-
craft, including USAF F-16s and A-10s, RAF Harrier GR. Mk 7s,
French Jaguars and Super Etendards, Italian AMXs, and Canadian
CF-18s. A subsequent late-morning launch featured 32 aircraft,
consisting of RAF Tornado GR. Mk 1s, French Jaguars, and USAF
______________
90The 15,000-ft restriction was never done away with over Serbia and Montenegro,
however, and over Kosovo it was eased only for FACs and for some weapon deliveries
in selected circumstances.
91Hewson, “Allied Force, Part II,” p. 102.
92R. Jeffrey Smith and Dana Priest, “Yugoslavia Near Goals in Kosovo,” Washington
Post, May 11, 1999.
RAND MR1365-3.3
900
800 U.S. sorties

Non-U.S. sorties
700
600
Number of sorties
500
400
300
200
100
0
24
31
14
21
28
12
19
26
9
ril
ay
ne
ne
ch
ch
ril
ril
ril
ay
ay
ay
Ap
Ju
Ju
Ap
Ap
Ap
M
ar
ar
M
Figure 3.3—U.S. and Allied Sorties Flown
F-16s, followed by 30 F-15Es and 16 more later in the surge. A mid-

afternoon strike with 28 jets was then followed by 24 more, with the
day finally ending with a midnight package of 38 strikers, including
B-1Bs and B-52s.93
Three days later, General Jumper declared that NATO had achieved
de facto air superiority over Yugoslavia, enabling attacking aircraft to
“go anywhere we want in the country, any time,” even though the
skies were admittedly “still dangerous.”94 Not long thereafter, an
option became available to attack from the north with 24 F/A-18Ds
of Marine Air Group 31 deployed to Taszar, Hungary. That option
______________
93Hewson, “Allied Force, Part II,” p. 109.
94Ibid., p. 110.
promised to further isolate Yugoslavia, make it appear surrounded,

and force its remaining air defenses to work harder by having to look
in more than one direction rather than mainly toward a single attack
axis from the west. It also promised to avoid adding to the already
severe air traffic congestion over the Adriatic and in other western
approaches to Yugoslavia.
As allied air operations against VJ troops in the KEZ became more

aggressive, a clear preference for the USAF’s A-10 over the Army’s
AH-64 Apaches in Albania became evident because weather condi-
tions over Kosovo had improved by that time, rendering the Apache’s
under-the-weather capability no longer pertinent and because
enough of the Serb IADS had been deemed weakened or intimidated
to make it safer to operate the A-10s at lower altitudes. Moreover, the
Apaches were deemed to be more susceptible to AAA and infrared
SAM threats than were the faster and higher-flying A-10s. President
Clinton himself later reinforced those reservations when he com-
mented in mid-May that the risk to the Apache pilots remained too
great and that because of recent weather improvements, “most of
what the Apaches could do [could now] be done by the A-10s at less
risk.” 95
Later on in May, allied fighters and USAF heavy bombers committed

against suspected enemy troop positions in the KEZ were joined for
the first time by USAF AC-130 gunships, which offered an additional
standoff capability against enemy vehicles and other ground targets
with their accurate 40mm Bofors gun, 25mm Gatling gun, and
105mm howitzer. The AC-130, however, was only used over areas
where there was no known or suspected presence of operational
SAM batteries and always flew above the reach of IR SAMs and AAA.
When targets of opportunity presented themselves on rare occasions,
sensor platforms that detected ground vehicular movement would
pass the coordinates and target characterization information to the
______________
95 Robert Burns, “Use of Apache Copters Is Not Expected Soon,” Philadelphia In-
quirer, May 19, 1999. In what may have been intended as an attempt to lessen the
sting of this leadership ruling, one Army source suggested that sending the Apaches in
had been meant all along merely as a scare tactic to induce Milosevic to negotiate.
The source added that if they had really been intended to be used, the more modern
and capable Apache Longbows would have been deployed instead. “Obviously, it was
just for show, not for go.” Rowan Scarborough, “Apaches Were Sent to Scare Serbs,”
Washington Times, May 21, 1999.
EC-130 ABCCC, which, in turn, would vector NATO attack aircraft

into the appropriate kill box, first to confirm that the targets were
valid and then to engage them. The ABCCC also controlled the
ingress and egress of attacking fighters and maintained battlespace
deconfliction throughout ongoing operations.
Once Serbia’s air defenses became a less imminent threat, the air war
also saw a heightened use of B-52s, B-1s, and other aircraft carrying
unguided bombs.96 By the end of May, some 4,000 free-fall bombs,
around 30 percent of the total number of munitions expended alto-
gether, had been dropped on known or suspected VJ targets in the
KEZ. There was a momentary resurgence of Serb SAM activity later
that month, with more than 30 SAMs reportedly fired on May 27, the
greatest number launched any night in nearly a month.97 That
heightened activity was assessed as reflecting a determined last-ditch
Serb effort to down at least one more NATO aircraft. (An F-117 had
been shot down during the air war’s fourth night, and a USAF F-16
had later been downed on the night of May 2.)98
______________
96The Block D version of the B-1 employed in Allied Force was configured to carry the
GBU-31 joint direct attack munition (JDAM), but only the B-2 actually delivered that
still-scarce munition.
97It bears noting here that the highly effective GAU-8 30mm cannon carried by the
A-10 saw use only 156 times in Allied Force because of the extreme slant range that
was required by the 5,000-ft altitude restriction (comments on an earlier draft by Hq
USAFE/SA, April 6, 2001). At that range, the principal problem for today’s A-10 pilots
is not hitting the target; it is seeing the target. At a 30-degree dive angle from 5,000 ft,
the slant range to target is 10,000 ft.
98 Glenn Burkins, “Serbs Intensify Effort to Down Allied Warplanes,” Wall Street
Journal, May 28, 1999. In the second instance, the ABCCC drew on instantly accessi-
ble satellite photos and maps maintained in a National Imagery and Mapping Agency
computerized database to identify potential obstacles, such as power lines, in order to
plot a safe course for the rescue helicopter that recovered the downed pilot. Bill Gertz
and Rowan Scarborough, “Inside the Ring,” Washington Times, May 19, 2000. Al-
though there was definitely a pronounced increase in enemy SAM activity during the
night of May 27 in an apparent effort to down a NATO pilot at any cost, it bears
stressing that there were no nights during Allied Force without at least a few SAM
shots, approximately 35 nights with 10 or more shots, and at least 13 nights with 20 or
more shots. The highest number of shots observed (significantly higher than the
number observed on May 27) was on the night of the F-16 loss. Overall, enemy SAM
activity levels tracked closely with allied air attack levels. Low-observable and cruise-
missile-only strikes prompted little enemy IADS reaction, whereas trolling for SAMs
with F-16CJs and CGs and large conventional attack packages always generated a pro-
portionately large enemy reaction. This trend remained consistent throughout the air
war from start to finish. Comments on an earlier draft by Hq USAFE/IN, May 18, 2001.
In what was initially thought to have been a pivotal turn of events in

the air effort against enemy ground forces, the newly enlarged and
hastily trained KLA, estimated to have been equipped with up to
30,000 automatic weapons, including heavy machine guns, sniper ri-
fles, rocket-propelled grenades, and antitank weapons, launched a
counteroffensive on May 26 against VJ troops in Kosovo. That thrust,
called Operation Arrow, involved more than 4,000 guerrillas of the
137th and 138th Brigades and drew artillery support from the Alba-
nian army, with the aim of driving into Kosovo from two points along
the province’s southwestern border, seizing control of the highway
connecting Prizren and Pec, and securing a safe route for the KLA to
resupply its beleaguered fighters inside Kosovo.
Operation Arrow represented the first major assault by KLA rebels in

more than a year. It was evidently intended to demonstrate both to
Milosevic and to NATO that the KLA remained a credible fighting
presence in Kosovo. The assault was thwarted at first by VJ artillery
and infantry counterattacks, which indicated that the VJ still had
plenty of fight in it despite 70 days of intermittent NATO bombing.
Three days after launching their assault, the rebels found themselves
badly on the defensive, with some 250 KLA fighters pinned down by
700 VJ troops near Mount Pastrik, a 6,523-ft peak just inside the
Kosovo-Albanian border.
For abundant good reasons, not least of which was a determination

to avoid even a hint of appearing to legitimize the KLA’s independent
actions, NATO had no interest in serving as the KLA’s de facto air
force and repeatedly refused to provide it with the equipment it
would have needed for its troops to have performed directly as
ground forward air controllers (FACs). The KLA did, however, receive
allied support in other ways. There had been earlier unconfirmed
reports going as far back as the air war’s second week that KLA
guerrillas had been covertly assisting NATO in the latter’s effort to
find and target VJ forces in Kosovo. 99 The first known direct NATO
air support to the KLA occurred on the day that Operation Arrow
commenced. It was confirmed both by KLA fighters in Albania and
by military officials in Washington.
______________
99Alessandra Stanley, “Albanian Fighters Say They Aid NATO in Spotting Serb Tar-
gets,” New York Times, April 2, 1999.
Although the Clinton administration denied helping the KLA directly,

U.S. officials did admit that NATO had responded to “urgent” KLA
requests for air support to turn back the VJ counterattack against its
embattled troops near Mount Pastrik. In addition to the support
they attempted to provide at Mount Pastrik, NATO aircraft attacked
VJ targets near the Kosovar villages of Bucane and Ljumbarda, en-
abling the rebels to capture those villages. The KLA kept NATO
informed of its positions in part so that its troops would not be inad-
vertently bombed, which had occurred two weeks earlier in an acci-
dental NATO attack on a KLA barracks in Kosari.100 KLA guerrillas
used cell phones to convey target coordinates to their base com-
manders, who, in turn, relayed that information to NATO military
authorities.
Throughout most of Allied Force, NATO and the KLA fought parallel
but separate wars against VJ and MUP forces in Kosovo, and both the
U.S. government and the KLA denied coordinating their operations
in advance. NATO did acknowledge, however, that rebel attacks on
the ground had helped flush out VJ troops and armor and to expose
them to allied air strikes on at least a few occasions, and that Clark
had authorized the communication of KLA target location informa-
tion to attacking NATO aircrews indirectly through the ABCCC. The
KLA further acknowledged that NATO air strikes had helped its
ground operations.101 Despite NATO denials throughout the air war
that it was aiding the KLA, it became evident that cooperation be-
tween the two was considerably greater than had been previously
admitted. As reported by KLA soldiers, the KLA had begun as early as
May 10 to supply NATO with target intelligence and other battlefield
information at NATO’s request, with the KLA’s chief of staff, Agim
Ceku, working with NATO officers in northern Albania. While refus-
ing to elaborate on specifics, KLA spokesmen admitted that Ceku had
been the KLA’s principal point of contact with NATO. It was also
Ceku who had participated in Croatia’s 1995 Operation Storm offen-
______________
100 Dana Priest and Peter Finn, “NATO Gives Air Support to Kosovo Guerrillas,”
Washington Post, June 2, 1999.
101 Marjorie Miller, “KLA Vows to Disarm If NATO Occupies Kosovo,” Los Angeles
Times, June 7, 1999.
sive that drove out the Krajina Serbs and helped end the fighting in
Bosnia. 102
Ultimately, VJ forces managed to repulse the KLA assault at Mount

Pastrik. To do so, however, they had to come out of hiding and move
in organized groups, making themselves potential targets, especially
for A-10s, on those infrequent occasions when they were detected
and approved for attack by the ABCCC or the CAOC.103 When KLA
actions forced VJ troops to concentrate enough tanks and artillery to
defend themselves, NATO aircraft were occasionally able to detect
and engage them. Enemy ground movements during the final two
weeks were often first noted by the E-8 joint surveillance target at-
tack radar system (Joint STARS) or other sensors, even though the VJ
studiously sought to maneuver in small enough numbers to avoid
being detected. The sensor operators would then transmit the coor-
dinates of suspected enemy troop concentrations to airborne for-
ward air controllers who, in turn, directed both unmanned aerial
vehicles (UAVs) and fighters in for closer looks, and ultimately for
attacks. 104 KLA ground movements were also displayed aboard the
ABCCC, which was coordinating and controlling NATO attacks
against VJ armored vehicles in Kosovo, deconflicting the attacking
aircraft, and ensuring that KLA forces in close contact with the VJ
were not inadvertently hit. Those operations represented classic in-
stances of close air support, with KLA and enemy forces in close
contact on the ground. The ABCCC and attacking NATO aircrews re-
ceived commands directly from the allied CAOC in Vicenza, Italy,
which, in at least one case, aborted an attack out of concern for hit-
ting KLA fighters. 105
Despite this heightened activity in the KEZ during the air war’s final
days, however, the attacks did better at keeping VJ and MUP troops
______________
102 Matthew Kaminski and John Reed, “NATO Link to KLA Rebels May Have Helped
Seal Victory,” Wall Street Journal, July 6, 1999.
103 William Drozdiak and Anne Swardson, “Military, Diplomatic Offensives Bring
About Accord,” Washington Post, June 4, 1999.
104 Tony Capaccio, “JSTARS Led Most Lethal Attacks on Serbs,” Defense Week, July 6,
1999, p. 13.
105 Michael R. Gordon, “A War out of the Night Sky: 10 Hours with a Battle Team,”
New York Times, June 3, 1999.
dispersed and hidden than they did at actually engaging and killing
them in any significant numbers. Most attack sorties tasked to the
KEZ did not release their weapons against valid military targets, but
rather against so-called dump sites for jettisoning previously unex-
pended munitions, sites that were conveniently billed by NATO tar-
get planners as “assembly areas.” Even the B-52s and B-1s, for all the
free-fall Mk 82 bombs they dropped during the final days, were
tasked with delivering a high volume of munitions without causing
any collateral damage. After the air war ended, it was never estab-
lished that any of the bombs delivered by the B-52s and B-1s had
achieved any militarily significant destructive effects, or that NATO’s
cooperation with the KLA had yielded any results of real operational
value. The steadily escalating attacks against infrastructure targets in
and around Belgrade that were taking place at the same time, how-
ever, were beginning to produce a very different effect on Serb be-
havior.
THE ENDGAME
On June 2, with Operation Allied Force working at peak intensity and
with weather and visibility for NATO aircrews steadily improving,
Russia’s envoy to the Balkans, former Prime Minister Viktor Cher-
nomyrdin, and Finland’s President Martti Ahtisaari, the European
Union representative, flew to Belgrade to offer Milosevic a plan to
bring the conflict to a close. Ahtisaari’s inclusion in the process was
said by one informed observer to have grown out of a suggestion by
Chernomyrdin that value might be gained from including a re-
spected non-NATO player on his mission.106 The same day, after the
two emissaries had essentially served him with an ultimatum that
had been worked out and agreed to previously by the United States,
Russia, the European Union, and Ahtisaari, Milosevic accepted an
international peace proposal. Under the terms of the proposed
agreement, he would accede to NATO’s demands for a withdrawal of
all VJ, MUP, and Serb paramilitary forces from Kosovo; a NATO-led
security force in Kosovo; an unmolested return of the refugees to
______________
106 Comments on an earlier draft by Ivo Daalder of the Brookings Institution, Wash-
ington, D.C., March 15, 2001. Daalder previously served as director for European af-
fairs on the National Security Council staff in 1995 and 1996, where he was responsible
for coordinating U.S. policy for Bosnia.
their homes; and the creation of a self-rule regime for the ethnic Al-
banian majority that acknowledged Yugoslavia’s continued
sovereignty over Kosovo. NATO would continue bombing pending
the implementation of a military-to-military understanding that had
been worked out between NATO and Yugoslavia on the conditions of
Yugoslavia’s force withdrawal. The agreement, which came on the
72nd day of the air effort, was ratified the day after, on June 3, by the
Serb parliament and was rationalized by Milosevic’s Socialist Party of
Serbia on the ground that it meant “peace and a halt to the evil
bombing of our nation.” 107
Milosevic later met with loyalist and opposition leaders to explain

the reasons for his decision to accept the peace plan. That was as
strong an indicator as any to date that the United States and NATO
were at the brink of success in their effort to get Yugoslavia’s 40,000
troops removed from Kosovo, the Kosovar refugees returned to their
homes, and NATO-dominated peacekeepers on Kosovo’s soil to en-
sure that the agreement was honored by Milosevic. The agreement
stipulated that once all occupying VJ and MUP personnel had de-
parted Kosovo, an agreed-upon contingent of Serbs—numbering
only in the hundreds, not thousands—could return to Kosovo to
provide liaison to the various peacekeeping entities commanded by
British Army Lieutenant General Sir Michael Jackson, help clear the
minefields that they had earlier laid, and protect Serb interests at re-
ligious sites and border crossings.
The two-page draft agreement further called for removing all Serb air
defense equipment and weapons deployed within 15 miles of the
Kosovo border by the first 48 hours so that NATO aircraft could verify
the troop withdrawals unmolested by any threats. The plan envis-
aged a U.S. sector to be controlled, first, by 1,900 Marines with light
vehicles and helicopters standing by aboard three ships in the
Aegean and, later, by the full American force complement made up
largely of Army tank and infantry units to be brought in from Ger-
many. U.S. forces, including the Marines from the 26th Marine Ex-
peditionary Unit at sea and three Army battalions from the 1st In-
fantry Division in Germany, would make up 15 percent of the overall
______________
107 Daniel Williams and Bradley Graham, “Yugoslavs Yield to NATO Terms,” Wash-
ington Post, June 4, 1999.
Kosovo Force (KFOR). The agreement similarly provided for British,

French, Italian, and German sectors.
NATO refused to commit itself to an early halt to its air attacks, since
its leaders knew that it would be extremely difficult to resume the
bombing once the refugees began coming home. During the negoti-
ations over the terms of Serb withdrawal, however, NATO pilots were
under orders not to attack any enemy positions unless in direct re-
sponse to hostile acts. After the Serb parliament agreed to the cease-
fire, no bombs fell on Belgrade for three consecutive nights. B-52
strikes against dispersed VJ forces, however, continued.
No sooner had this accord been reached in principle than NATO and
Serb military officials failed to reach an understanding on the condi-
tions for VJ and MUP withdrawal. The talks quickly degenerated into
haggling over when NATO would halt its air attacks and whether
Serbia would have more than a week to get its troops out of Kosovo.
The proximate cause of the breakdown in talks was a Serb demand
that the UN Security Council approve an international peacekeeping
force before NATO troops entered Kosovo. That heel-dragging sug-
gested that the Serbs were seeking to soften some of the terms of the
settlement or, perhaps, were looking for more time to continue their
fight with the KLA. Secretary Cohen and General Shelton allowed
that extending the Yugoslav withdrawal by several days would be ac-
ceptable but that they would not countenance any deliberate at-
tempts at delay. More specifically, the implementation of the Serb
withdrawal was hung up on differences over the sequencing of four
events: the start of the enemy pullout, a pause in NATO bombing,
the passage of a UN resolution, and the entry of international peace-
keepers with a “substantial NATO content.” In response to this will-
ful foot-dragging, NATO’s attacks, which initially had been scaled
back after Milosevic accepted the proposed peace plan, resumed
their previous level of intensity.
On June 7, at the same time as the talks were under way, VJ forces
launched a renewed counterattack against the KLA in an area south
of Mount Pastrik, where the two sides had been locked in an artillery
duel since May 26. For a time, a major breakthrough in NATO’s air
effort was thought to have occurred when the defending KLA forces
flushed out VJ troops who had been dispersed around Mount Pastrik,
creating what NATO characterized as a casebook target-rich envi-
ronment. Thanks to improved weather, a noticeable degradation of

Serb air defenses, and the effective role thought to have been played
by the KLA in forcing VJ troops to come out of hiding, two B-52s and
two B-1Bs dropped a total of 86 Mk 82s on an open field in a daytime
raid near the Kosovo-Albanian border where VJ forces were believed
to have been massed.108 The initially estimated number of enemy
troops caught in the open by the attack was 800 to 1,200, with early
assessments suggesting that fewer than half had survived the at-
tack. 109 It later appeared, however, that the number of enemy ca-
sualties was considerably less than originally believed—if, indeed,
the attacking bombers had killed a significant number of VJ troops
at all.
Whatever the case, the following day the United States, Russia, and
six other member-states agreed on a draft UN Security Council reso-
lution to end the conflict. The resolution called for a complete with-
drawal of Serb troops, police, and paramilitary forces from Kosovo
and for all countries to cooperate with the war crimes tribunal that
had indicted Milosevic. 110 The sequence finally agreed to was that
the Serb force withdrawal would commence, NATO would concur-
rently halt its bombing, and only after those two actions occurred
would the Security Council vote on the text of the agreement. The
last provision was a token concession to Russia and China, whose
representatives had insisted that the bombing be stopped before any
Security Council vote was taken.
In the end, the VJ acceded to a six-page agreement that permitted a

KFOR presence of 50,000 peacekeepers commanded by a NATO gen-
eral and having sweeping occupation powers over Kosovo. By the
terms of the agreement, Serb forces would withdraw along four des-
ignated routes over 11 days, under the constant threat of resumed
bombing in case of any willful delays.111 (Belgrade had asked for 17
______________
108 R. Jeffrey Smith and Molly Moore, “Plan for Kosovo Pullout Signed,” Washington
Post, June 10, 1999.
109 William Drozdiak, “Yugoslav Troops Devastated by Attack,” Washington Post, June
9, 1999.
110 Smith and Moore, “Plan for Kosovo Pullout Signed.”
111 At one point in the negotiations, the VJ military delegation leader, Colonel General
Svetozar Marjanovic, abruptly walked out of the talks, stating that he needed to
days.) Kosovo was to be ringed by a 5-km buffer zone, and NATO

was to provide for the safe return of all refugees. After 78 days of
continual bombing by NATO, the agreement was finally signed on
June 9 in a portable hangar at a NATO airfield in Kumanovo, Mace-
donia, five miles south of the Yugoslav border. The 11 days granted
to Yugoslavia for the troop withdrawal was another diplomatic con-
cession, considering that NATO had initially insisted that the with-
drawal be completed in 7 days.
NATO finally stopped the bombing upon verifying that the Serb
withdrawal had begun, after which the UN Security Council ap-
proved, by a 14-0 vote with China abstaining, a resolution putting
Kosovo under international civilian control and the peacekeeping
force under UN authority. With that, President Clinton declared that
NATO had “achieved a victory.”112
Once NATO peacekeeping forces moved in on the ground in Kosovo,

they began discovering the full extent of Serb atrocities committed
against the Kosovar Albanians. Among other things, they found an
interrogation center in Pristina that had been used by Serb police, in
which thousands of Kosovar suspects were said to have been
“processed.” Inside the bowels of the building, they came across
garrotes with wooden handles, brass knuckles, broken baseball bats,
chainsaws, and leather manacles and straps. They also were told by
surviving Kosovars that the Serb police had spent three days burning
records before the British paratroopers finally arrived. 113 Later, the
British government estimated that some 10,000 ethnic Albanians had
died at the hands of marauding Serbs during the course of Operation
Allied Force.114
As the last of some 40,000 VJ and MUP personnel exited Kosovo on

June 20 a few hours ahead of NATO’s deadline, NATO declared a
formal end to the air war. The bombing had earlier been suspended
_____________________________________________________________
“consult with authorities in Belgrade.” He made it only to a border post and returned
to the negotiating table within an hour.
112 Tim Weiner, “From President, Victory Speech and a Warning,” New York Times,
June 11, 1999.
113 See Julian Barger, “Bloody Paper Chain May Link Torture to Milosevic,” The
Guardian, June 18, 1999.
114 Ian Black and John Hooper, “Serb Savagery Exposed,” The Guardian, June 18, 1999.
informally for 10 days when the first Serb troops began leaving
Kosovo. The departure of the last Serb forces and the arrival of the
KFOR peacekeepers effectively brought an end to Yugoslav control
over a province that had been a special and even sacred preserve of
Serbia for centuries.
Initial estimates just before the cease-fire went into effect claimed
that the air war had taken out 9 percent of Serbia’s soldiers (10,000 of
114,000), 42 percent of its aircraft (more than 100 of 240), 25 percent
of its armored fighting vehicles (203 of 825), 22 percent of its artillery
pieces (314 of 1,400), and 9 percent of its tanks (120 of 1,270).115
After the cease-fire, the Pentagon claimed that the operation had de-
stroyed 450 enemy artillery pieces, 220 armored personnel carriers,
120 tanks, more than half of Yugoslavia’s military industry, and 35
percent of its electrical power-generating capacity.116 General Shel-
ton reported that 60 percent of the infrastructure of the Yugoslav 3rd
Army, the main occupying force in Kosovo, had been destroyed,
along with 35 percent of the 1st Army’s infrastructure and 20 percent
of the 2nd Army’s.117 The U.S. Air Force’s deputy chief of staff for air
and space operations, Lieutenant General Marvin Esmond, an-
nounced that the allied bombing effort had destroyed a presumed 80
percent of Yugoslavia’s fixed-wing air force, zeroed out its oil refining
capability, and eliminated 40 percent of its army’s fuel inventory and
40 percent of its ability to produce ammunition. Many of these ini-
tial assessments were later discovered to have been overdrawn by a
considerable margin.
In the final tally, allied aircrews flew 38,004 out of a planned 45,935
sorties in all, of which 10,484 out of a planned 14,112 were strike sor-
ties.118 A later report to Congress by Secretary Cohen and General
Shelton claimed that more than 23,300 combat missions, including
______________
115 Michael R. Gordon and Eric Schmitt, “Shift in Targets Let NATO Jets Tip the Bal-
ance,” New York Times, June 5, 1999.
116 Weiner, “From President, Victory Speech and a Warning.”
117 Bradley Graham, “Air Power ‘Effective, Successful,’ Cohen Says,” Washington Post,
June 11, 1999.
118 Operation Allied Force and Operation Joint Guardian briefing charts dated August
19, 1999, provided to the author by Air Marshal Sir John Day, RAF, UK Ministry of De-
fense director of operations in Allied Force, RAF Innsworth, United Kingdom, July 25,
2000.
defensive counterair patrols and defense suppression attacks, were

flown altogether, entailing weapon releases against roughly 7,600
desired mean points of impact (DMPIs) on fixed targets and slightly
more than 3,400 presumed mobile targets of opportunity.119 As for
the air war’s intensity over time, what started out as little more
than 200 combat and combat-support sorties a day eventually rose to
over 1,000 sorties a day by the time of the cease-fire.120 All told, 28
percent of the sorties flown were devoted to direct attack, with 12
percent going to SEAD, 13 percent to attacks against dispersed en-
emy forces in Kosovo, 16 percent to defensive counterair patrols, 20
percent to inflight refueling, and 11 percent to other combat support
missions (including AWACS, Joint STARS, ABCCC, EC-130 jammers,
airlift, and combat search and rescue).121 Figure 3.4 shows the
breakout of U.S. sorties flown by aircraft type.
According to the final air operations database later compiled by Hq

USAFE, 421 fixed targets in 11 categories were attacked over the 78-
day course of Allied Force, of which 35 percent were believed to have
been destroyed, with another 10 percent sustaining no damage and
the remainder suffering varying degrees of damage from light to se-
vere. The largest single fixed-target category entailed ground-force
facilities (106 targets), followed by command and control facilities
(88 targets) and lines of communication, mostly bridges (68 targets).
Other target categories included POL-related facilities (30 targets),
industry (17 targets), airfields (8 targets), border posts (18 targets),
and electrical power facilities (19 targets). In addition, 7 so-called
counterregime targets were assessed as having sustained overall light
damage. Finally, 60 targets were associated with Serb air defenses in
two categories (radars and launch equipment), out of which two of
______________
119 Secretary of Defense William S. Cohen and Chairman of the Joint Chiefs of Staff
General Henry H. Shelton, Kosovo/Operation Allied Force After-Action Report,
Washington, D.C., Department of Defense, Report to Congress, January 31, 2000, p. 87.
120 Ibid., p. 68.
121 Operation Allied Force and Operation Joint Guardian briefing charts dated August
19, 1999.
RAND MR1365-3.4
Total: 30,018
UAVs
(496)
Fighter
(8,889)
Intratheater
airlift
(11,480)
Bomber
(322)
Special Tanker
Ops (6,959)
ISR
(834) (1,038)
Figure 3.4—USAF Sortie Breakdown by Aircraft Type
three SA-2s, 11 of 16 SA-3s, and 3 of 25 STRAIGHT FLUSH radars as-

sociated with the SA-6 were assessed as having been destroyed. 122
______________
122 “AWOS Fact Sheet,” Hq USAFE/SA, December 17, 1999. See also William M. Arkin,
“Top Air Force Leaders to Get Briefed on Serbia Air War Report,” Defense Daily, June
13, 2000, p. 1. As attested by cockpit display videotapes released to the press
throughout the air war, allied air attacks succeeded in taking out quite a few more SA-6
launchers than those accounted for here. However, since the STRAIGHT FLUSH radar
formed the core of an SA-6 battery, the battery was considered operational until the
STRAIGHT FLUSH was destroyed. Comments on an earlier draft by Hq USAFE/IN,
May 18, 2001.
As for the 28,018 munitions (excluding TLAMs) that were expended

altogether, a full one-third were general-purpose Mk 82 unguided
bombs dropped by B-52s and B-1s during the war’s final two weeks.
Figure 3.5 presents the trend of U.S. and allied munitions expendi-
ture over the 78-day course of Allied Force. Figure 3.6 shows the
number of precision weapons and nonprecision weapons delivered
daily over the same period. Of that number, the United States deliv-
ered 83 percent, or all but 4,703 (see Figure 3.7).
In a telling reflection of the sparse intelligence available on the

location of enemy SAMs and of NATO’s determination to avoid los-
ing even a single aircrew member, some 35 percent of the overall
RAND MR1365-3.5
600
U.S.
500 Non-U.S.
400
Number of munitions
300
200
100
0
24
31
14
21
28
12
19
26
9
ril
ay
ne
ne
h
ril
ril
ril
ay
ay
ay
Ap
M
c
Ju
Ju
Ap
Ap
Ap
M
ar
ar
M
Figure 3.5—U.S. and Allied Ground-Attack Munitions Expended

(excluding TLAM)
RAND MR1365-3.6
1600
Non-PGM
1400 PGM
1200
Number of munitions
1000
800
600
400
200
0
24
31
14
21
28
12
19
26
9
ril
ay
ne
ne
ch
ch
ril
ril
ril
ay
ay
ay
Ap
Ju
Ju
Ap
Ap
Ap
M
ar
ar
M
Figure 3.6—U.S. and Allied Munitions Expenditures by Type
effort (including both direct attack and mission support) was

directed against enemy air defenses. Thanks in part to the weight of
that effort, only two allied aircraft were downed and not a single
friendly fatality was incurred, save for two AH-64 pilots who were
killed in a training accident in Albania (see Chapter Five for more on
these incidents). Even at that, however, enemy SAMs were effectively
suppressed but not often destroyed.
RAND MR1365-3.7
Total: 28,018
Allied
(4,703)
17%
83%
U.S.
(23,315)
Figure 3.7—Total Numbers of Munitions Expended

Stealth bomber. Used for the first time in combat during Allied Force after having been in line serv-
ice since 1993, a USAF B-2 prepares to take on fuel midway across the Atlantic Ocean during one of
two preplanned tanker hookups en route to target. The low-observable bomber, operating nonstop
from its home base in the United States, was the first allied aircraft to penetrate Serb defenses on
opening night.
Safe recovery. An F-117 stealth attack aircraft lands at Aviano Air Base, Italy, just after sunrise fol-
lowing a night mission into the most heavily defended portions of Serbia. During the air war’s
fourth night, an F-117 was downed just northwest of Belgrade, most likely by a lucky SA-3 shot, in
the first-ever loss of a stealth aircraft in combat. (The pilot was promptly retrieved by CSAR forces.)
Heavy players. A venerable USAF B-52H bomber stands parked on the ramp at RAF Fairford,
England, as a successor-generation B-1B takes off on a mission to deliver as many as 80 500-lb Mk
82 bombs or 30 CBU-87 cluster bomb units against enemy barracks and other area targets. An
AGM-86C CALCM fired from standoff range by a B-52 was the first allied weapon to be launched in
the war.
Final checks. Two Block 40 F-16CGs from the 555th Fighter Squadron at Aviano taxi into the arm-
ing area just short of the runway for one last look by maintenance technicians before taking off on
a day mission to drop 500-lb GBU-12 laser-guided bombs on “flex” targets of opportunity in Serbia
or Kosovo, as directed by airborne FACs and as approved, in some cases, by the CAOC.
Burner takeoff. An F-15E from the 494th Fighter Squadron home-based at RAF Lakenheath,
England, clears the runway at Aviano in full afterburner, with CBU-87 cluster munitions shown
mounted on its aftmost semiconformal fuselage weapons stations. Eventually, some F-15E strike
sorties into Serbia and Kosovo were flown nonstop to target and back directly from Lakenheath.
On the cat. A U.S. Navy F/A-18C assigned to Fighter/Attack Squadron 15 is readied for a catapult
launch for an Allied Force day combat mission from the aircraft carrier USS Theodore Roosevelt
cruising on station in the Adriatic Sea. On April 15, carrier-based F/A-18s figured prominently in a
major CAOC-directed air strike on the Serb air base at Podgorica, Montenegro.
SAM hunter. This Block 50 F-16CJ in the arming area at Aviano shows an AGM-88 high-speed anti-
radiation missile (HARM) mounted on the left intermediate wing weapons station, with an AIM-
9M air-to-air missile on the outboard station and an AIM-120 AMRAAM on the wingtip missile rail.
The USAF’s F-16CJ inventory was stressed to the limit to meet the SEAD demand of Allied Force.
Combat support. U.S. Navy and Marine Corps EA-6B Prowler electronic warfare aircraft, like this
one shown taxiing for takeoff at Aviano, provided extensive and indispensable standoff jamming of
enemy early warning and IADS fire-control radars to help ensure unmolested allied strike opera-
tions, including B-2 and F-117 stealth operations, against the most heavily defended enemy targets
in Serbia.
Task Force Hawk. A U.S. Army AH-64 Apache attack helicopter flares for landing at the Rinas air-
port near Tirana, Albania, following a ferry flight from its home base at Illesheim, Germany. In all,
24 Apaches were dispatched to Albania with the intent to be used in Operation Allied Force, but
none saw combat in the end because of concerns for the aircraft’s prospects for survival in hostile
airspace.
Cramped spaces. This USAF C-17 parked on the narrow ramp at Rinas airport, incapable of
accommodating the larger C-5, was one of many such aircraft which provided dedicated mobility
service to TF Hawk. In more than 500 direct-delivery lift sorties altogether, C-17s moved 200,000-
plus short tons of equipment and supplies to support the Army’s deployment within the span of
just a month.
Eagle eye. Ground crewmen at RAF Lakenheath prepare a LANTIRN targeting pod to be mounted
on an F-15E multirole, all-weather fighter. The pod, also carried by the Navy F-14D and the USAF
F-16CG, contains a forward-looking infrared (FLIR) sensor for target identification at standoff ranges
day or night, as well as a self-contained laser designator for enabling precision delivery of LGBs.
Deadly force. Munitions technicians at RAF Fairford prepare a CBU-87 cluster bomb unit for load-
ing into a USAF B-1B bomber in preparation for a mission against fielded Serbian forces operating
in Kosovo. With a loadout of 30 CBU-87s—more than five times the payload of an
F-15E—the B-1 can fly at fighter-equivalent speeds more than 4,200 nautical miles unrefueled.
Help from an ally. One of 18 CF-18 Hornet multirole fighters deployed in support of Allied Force
from Canadian Forces Base Cold Lake, Alberta, Canada, is parked in front of a hardened aircraft
shelter at Aviano. The aircraft mounts two 500-lb GBU-12 laser-guided bombs on the outboard
wing pylons and two AIM-9M air-to-air missiles on the wingtip rails.
Force protection. A security guard stands watch over a USAF C-17 airlifter at the Rinas airport in
support of the U.S. Army’s Apache attack helicopter deployment to Albania. For a time, reported
differences between on-scene Air Force and Army commanders with respect to who was ultimate-
ly responsible for the airfield made for discomfiting friction within the U.S. contingent.
Flexing into the KEZ. An AGM-65 Maverick-equipped A-10 from the USAF’s 52nd Fighter Wing
stationed at Spangdahlem Air Base, Germany, takes off from Gioia del Colle to provide an on-call
capability against possible Serb targets detected in Kosovo by allied sensors, including the TPQ-36
and TPQ-37 firefinder radars operated by the U.S. Army on the high ground above Tirana, Albania.
Round-the-clock operations. An F-16 pilot readies himself for a night mission over Serbia, his hel-
met shown fitted with a mount for night-vision goggles. Used in conjunction with compatible
cockpit lighting, NVGs made possible night tactics applications, including multiaircraft formations
and simultaneous bomb deliveries, which otherwise could only have been conducted during day-
light.
Night refueling. A USAF F-15C air combat fighter, shown here through a night-vision lens, moves
into the precontact position to take on fuel from a KC-135 tanker before resuming its station to
provide offensive counterair protection for attacking NATO strikers. With a loss of six MiGs in aer-
ial combat encounters the first week, Serb fighters rarely rose thereafter to challenge NATO’s con-
trol of the air.
Splash one Fulcrum. A team of U.S. military personnel examines the remains of an enemy MiG-
29 fighter (NATO code name Fulcrum) which was shot down in Bosnian airspace by a USAF F-15C
on the afternoon of March 26, 1999. The downed aircraft, which appeared to have strayed from its
planned course due to a loss of situation awareness by its pilot, brought to five the number of
MiG-29s destroyed in early Allied Force air encounters.
Hard-target killer. This F-15E pilot, a USAF major assigned to the 494th Fighter Squadron, looks
over a 4,700-lb electro-optically-guided GBU-28 bunker-buster munition mounted on his aircraft’s
centerline stores station. The aircraft, one of a two-ship flight of F-15Es (call sign Lance 31 and 32),
delivered the weapon on April 28, 1999, against an underground hangar at the Serb air base at
Podgorica.
Precision attack. In April 1999, a single B-2 achieved six accurately placed GBU-31 JDAM hits
against six runway-taxiway intersections at the Obvra military airfield in Serbia, precluding opera-
tions by enemy fighters until repairs could be completed. This post-strike image graphically shows
the B-2’s ability with JDAM to achieve the effects of mass without having to mass, regardless of
weather.
A bridge no more. Another post-strike battle-damage assessment image shows this bridge in
Serbia cut in two places by a precision bombing attack. Sometimes enemy bridges were dropped
at the behest of NATO target planners to prevent the flow of traffic over them. At other times, they
were attacked and damaged to sever key fiber-optic communications lines that were known to run
through them.
Before and after. This bridge over the Danube River near Novi Sad in Serbia, shown here in both
pre- and post-strike imagery, was all but completely demolished by precision bombing on Day 9 of
Allied Force as Phase III of the air war, for the first time, ramped up operations to include attacks
against not only Serbian IADS and fielded military assets but also key infrastructure targets.
Effects-based targeting. For three consecutive nights beginning on May 24, U.S. aircraft struck
electrical power facilities in Belgrade, Novi Sad, and Nis, the three largest cities in Serbia, shutting
off electrical power to 80 percent of Serbia. This transformer yard in Belgrade was one such target
that was attacked in what was arguably the most influential strike of Allied Force to that point.
Chapter Four
WHY MILOSEVIC GAVE UP WHEN HE DID
As might have been predicted, disagreements arose after the cease-

fire went into effect over which of the air war’s target priorities
(fielded forces or infrastructure assets) was more crucial to produc-
ing the outcome. Contention also arose over the more basic question
of the extent to which the air effort as a whole had been the cause of
Milosevic’s capitulation. On the one hand, there was the view of
those air power proponents who were wont to conclude up front that
“for the first time in history, the application of air power alone forced
the wholesale withdrawal of a military force from a disputed piece of
real estate.”1 On the other hand, there was the more skeptical view
offered by the commander of the international peacekeeping forces
in Kosovo, British Army Lieutenant General Sir Michael Jackson, who
suggested that “the event of June 3 [when the Russians backed the
West’s position and urged Milosevic to surrender] was the single
event that appeared to me to have the greatest significance in ending
the war.” Asked about the effects of the air attacks, Jackson, an
avowed critic of air power, replied tartly: “I wasn’t responsible for
the air campaign; you’re asking the wrong person.”2
______________
1John A. Tirpak, “Lessons Learned and Re-Learned,” Air Force Magazine, August 1999,
p. 23.
2Andrew Gilligan, “Russia, Not Bombs, Brought End to War in Kosovo, Says Jackson,”
London Sunday Telegraph, August 1, 1999. To his credit, Jackson did later testify to the
Commons Defense Committee of Britain’s parliament that “the effect of the strategic
bombing, I suspect, was much weightier than the damage being done to the [Serb]
army in Kosovo.” “General Admits NATO Exaggerated Bombing Success,” London
Times, May 11, 2000.
67
We may never know for sure what mix of pressures and inducements
ultimately led Milosevic to admit defeat, at least until key Serb
archival materials become available or those closest to Milosevic
during the air war become disposed to offer first-hand testimony.
Asked by a reporter why Milosevic folded if the bombing had not
defeated him militarily, Clark, who knew the Serb dictator well from
previous negotiating encounters, replied: “You’ll have to ask Milo-
sevic, and he’ll never tell you.”3 Yet why Milosevic gave in and why
he did so when he did are by far the most important questions about
the air war experience, since the answers, insofar as they are know-
able, will help to lay bare the coercive dynamic that ultimately swung
the outcome of Allied Force. It need hardly be said that such insight
can be of tremendous value in informing any strategy ultimately
chosen by the United States and its allies for future interventions of
that sort. Accordingly, it behooves analysts to make every effort to
delve further into this innermost mystery of the air war, since even
approximate answers, if buttressed by valid evidence, are almost
certain to be more useful to senior policymakers than most “lessons”
of a more technical nature regarding how specific systems worked
and how various procedural aspects of the operation could have
been handled better, important as the latter questions are.
In the search to understand what ultimately occasioned NATO’s
success, one can, of course, insist that air power alone was the cause
of Milosevic’s capitulation in the tautological sense that Allied Force
was an air-only operation and that in its absence, there would have
been no reason to believe that he would have acceded to NATO’s
demands.4 Yet as crucial as the 78-day bombing effort was in
bringing Milosevic to heel, there is ample reason to be wary of any
intimation that NATO’s use of air power produced that ending with-
out any significant contribution by other factors. On the contrary,
numerous considerations in addition to the direct effects of the
bombing in all likelihood interacted to produce the Serb dictator’s
eventual decision to cave in. As Ivo Daalder and Michael O’Hanlon
have remarked, in a balanced reflection on this point, “air power
______________
3 Michael Ignatieff, “The Virtual Commander: How NATO Invented a New Kind of
War,” The New Yorker, August 2, 1999, p. 31.
4 See, for example, Rebecca Grant, “Air Power Made It Work,” Air Force Magazine,
November 1999, pp. 30–37.
Why Milosevic Gave Up When He Did 69
might best be thought of as the force driving Milosevic into a dead-

end corner and threatening to crush him against the far wall. But
had NATO not remained unified, Russia not joined hands with NATO
in the diplomatic endgame, and the alliance not begun to develop a
credible threat of a ground invasion, Milosevic might have found
doors through which to escape from the corridor despite the aerial
punishment.”5
CONSIDERATIONS IN ADDITION TO THE BOMBING

Beyond the obvious damage that was being caused by NATO’s air
attacks and the equally obvious fact that NATO could have continued
bombing indefinitely and with virtual impunity, another likely factor
behind Milosevic’s capitulation was the fact that the sheer depravity
of Serbia’s conduct in Kosovo had stripped it of any remaining ves-
tige of international support—including, in the end, from its princi-
pal backers in Moscow. Although Milosevic’s loss of Russian support
may not have been the determining factor behind his capitulation, it
was, without question, a contributing factor. A high-level official in
the Clinton administration who was directly involved in setting poli-
cies for Operation Allied Force later commented that with respect to
the numerous ongoing diplomatic efforts to backstop the coercive
bombing, Russia was “a key arrow in the quiver.”6 That became
most clearly apparent when Russian President Boris Yeltsin called
Clinton on April 25, the last the day of the NATO summit, and, in an
unprecedentedly long 75-minute conversation, expressed his con-
cerns over the escalating air war and offered to send former Russian
Prime Minister Chernomyrdin as his personal envoy to help find a
negotiated solution. Once Milosevic came face to face with the real-
ization that Russia had joined the West in pressing for a settlement of
______________
5Ivo H. Daalder and Michael E. O’Hanlon, Winning Ugly: NATO’s War to Save Kosovo,
Washington, D.C., The Brookings Institution, 2000, p. 184.
6This official, in an interview with RAND staff members in Washington on June 11,
2000, further claimed that the White House was not surprised when Milosevic
accepted the deal on June 3, since the administration was confident that once Cher-
nomyrdin had agreed to NATO’s terms, it was merely a matter of time before a suc-
cessful denouement would be reached, considering that Chernomyrdin knew Milose-
vic’s bottom line and would not have signed up for any arrangement that he knew
Milosevic would not accept. What was surprising, the official said, was that Milosevic
did not first seek to buy time by proffering more “half-loaf” compromise deals.
the Kosovo standoff, he knew that he had lost any remaining trace of
international backing.
On top of that was the sense of walls closing in that Milosevic must
have had when he was indicted as a war criminal by a UN tribunal
only a week before his loss of Moscow’s support. On May 27, that
tribunal charged Milosevic and four of his senior aides—including
General Dragoljub Ojdanic, the Yugoslav army chief, and Vlajko Sto-
jilkovic, the interior minister responsible for the MUP—with crimes
against humanity for having deported more than 700,000 ethnic Al-
banians and having allegedly murdered 340 innocents, mostly young
men. Even if that indictment did not give Milosevic pause in and of
itself, it almost surely closed the door on any remaining chance that
Russia might change course and resume its support for him.
Yet a third factor, this one a direct second-order result of the bomb-
ing, may have been mounting elite pressure behind the scenes. As
the air attacks encroached more on Belgrade proper, Secretary Co-
hen reported that senior VJ leaders had begun sending their families
out of Yugoslavia, following a similar action earlier by members of
the Yugoslav political elite and reflecting possible concern among
top-echelon commanders that Milosevic had led them down a blind
alley in choosing to take on the United States and NATO.7 U.S. offi-
cials indicated that during the last week of the air war, VJ leaders had
swung from supporting Milosevic on Kosovo to openly rebelling and
pressuring the Serb dictator to agree to NATO’s terms. Cohen’s re-
port of increasing demoralization among the VJ’s most senior leaders
as they helplessly watched the escalating destruction all around
them gave rise to hopes within the Clinton administration that Milo-
sevic might be looking for a face-saving way out. 8 The fact that the
bombing effort caused more infrastructure damage during its last
week than during its entire first two months was thought by some to
have reawakened old tensions between Milosevic and an army lead-
ership that was said to have never fully trusted him.
______________
7Daniel Williams and Bradley Graham, “Milosevic Admits to Losses of Personnel,”
Washington Post, May 13, 1999.
8 Interview with Secretary of Defense William S. Cohen, “Milosevic Is Far Weaker
Now,” USA Today, May 14, 1999.
A related factor may have been mounting heat from Milosevic’s

cronies among the Yugoslav civilian oligarchy, prompted by the con-
tinued bombing of military-related industries, utilities, and other in-
frastructure targets in and around Belgrade in which they had an
economic stake and whose destruction increasingly threatened to
bankrupt them.9 On that point, administration officials remarked
that among other things, the dropping of bridges throughout Serbia
by NATO air attacks had hindered the activities of smugglers who
represented a key source of income for those cronies. Moreover, CIA
and other allied intelligence organizations were said to have been
gathering information on the bank accounts and business interests
of Milosevic and his closest partners, the latter of whom were starting
to pressure him to call it quits.10
Finally, U.S. psychological operations could have been a contributing

factor, although the evidence for that remains both spotty and less
than convincing. One report to that effect suggested that Milosevic’s
wife was becoming “increasingly hysterical” as the bombing intensi-
fied and that Milosevic himself was finally pushed over the edge after
the United States, via a “friendly intermediary,” shipped him a video-
tape showing what a fuel-air explosive could do to his forces—at
roughly the same time as the KLA’s counteroffensive in Kosovo
forced VJ troops into the open and exposed them to NATO fire.11
Apart from the fact that fuel-air explosives are not currently main-
tained in the U.S. munitions inventory, this claim presumed that the
______________
9Paul Richter, “Officials Say NATO Pounded Milosevic into Submission,” Los Angeles
Times, June 5, 1999. The possible effects of the bombing on what one might call sec-
ond-tier Serb leaders are especially noteworthy, in that they suggest that the elite sub-
structure of an enemy’s hierarchy may make for more lucrative leadership targets than
the “big guys.” Unlike the topmost political leaders, these second-tier individuals
have “retirement plans,” in that they have options to recoup their interests under a
new regime. They thus may be more malleable than their bosses, even as they are of-
ten critical to their bosses’ survival. I am grateful to Colonel Robert Owen, USAF, for
having suggested this intriguing idea to me.
les Times, June 9, 2000. A persistent concern that tended to inhibit a truly aggressive
use of such information entailed the liability implications of information attacks
against foreign bank accounts, as well as official worries about the Pandora’s box that
might be opened if the United States began playing that game, thus rendering its own
economy susceptible to similar measures in return.
11 Tom Walker, “Bomb Video Took Fight out of Milosevic,” London Sunday Times,
January 30, 2000.
VJ’s troops were a particularly valued asset for Milosevic, which, by

all indications, they were not.
THE PROSPECT OF A GROUND INVASION

Among the many considerations that converged to produce Milose-
vic’s eventual capitulation, the most discomfiting to him over the
long run—apart from the bombing itself—may well have been what
he perceived, rightly or wrongly, to be the prospect of an eventual
NATO ground intervention. Whatever NATO’s declared stance on
the ground-war issue may have been, its actions as the air war pro-
gressed spoke louder than its words.
To begin with, Operation Allied Harbor, set in motion as early as

April 8, aimed at putting some 8,000 NATO ground troops into
Macedonia to help with refugee aid efforts. More significantly, a
32,000-person NATO Stabilization Force (soon to number 50,000) pa-
trolling Bosnia-Herzegovina, and 7,500 additional NATO troops in
Albania deployed to perform humanitarian work there made for an
undeniable signal that a NATO ground presence was forming in the
theater. That presence included 2,400 combat-ready U.S. Marines
aboard three warships in the Adriatic to provide force protection for
the Marine F/A-18s that were operating out of the former Warsaw
Pact air base at Taszar. In addition, some 5,000 U.S. Army troops,
with a substantial artillery and armor complement, accompanied the
24 AH-64 Apache helicopters that were sent to Albania in late April.
There is every reason to believe that this deployment, along with
NATO’s subsequent decision to enlarge the Kosovo peacekeeping
force (KFOR) to as many as 50,000 troops, was assessed by Milosevic
as an indication that a NATO ground option was at least being kept
open.
Taking advantage of a covert relationship between the CIA and the

KLA, NATO also had begun probing the capability and extent of the
VJ’s ground defenses, an inquiry that most likely did not escape
Milosevic’s attention. In a related development, NATO engineers on
May 31 began widening and reinforcing a key access road from
Durres to Kukes on the Kosovo-Albanian border so that it could sup-
port the weight of a main battle tank. Earlier, Clark had authorized
the engineers to strengthen the road to handle refugee traffic only,
but they made it strong enough to support the Bradley armored
fighting vehicle (AFV). This time, only three days before Milosevic fi-
nally called it quits, Washington gave Clark permission to send in
another engineering battalion to make the road capable of support-
ing M1A2 Abrams tanks and artillery. 12
Beyond that, Milosevic may have gotten wind of a secret NATO plan
for a massive ground invasion code-named Plan B-minus, which was
slated to be launched the first week of September if approved by
NATO’s political leaders. In support of this plan, Britain had agreed
to contribute the largest single national component up to that time
(50,000 troops) to an envisaged 170,000-man contingent; the United
States would have contributed at least 100,000 more. Developed by a
secret planning team at NATO’s military headquarters in Mons, Bel-
gium, Plan B-minus relied heavily on previous plans going back to
June 12, 1998, which featured six land-attack options, including a full
invasion of Serbia itself (Plan Bravo, with 300,000 NATO troops). The
chief of Britain’s defense staff, General Sir Charles Guthrie, later con-
firmed the outlines of this plan. 13 Milosevic was said by a well-
placed NATO source to have been at least broadly informed of NATO
thinking with respect to it. Indeed, as the UK Ministry of Defense’s
director of operations in Allied Force, RAF Air Marshal Sir John Day,
later commented, “the decision to increase KFOR was militarily right
in itself, but it was also a form of heavy breathing on Milosevic and a
subtle way of moving to B-minus while keeping the coalition to-
gether. The move also had the effect of shortening our timelines for
B-minus. It is true that the forces that were being prepared for
KFOR-plus were the core elements of what would then have become
B-minus, the full ground invasion.”14
In a sign that such indicators may have begun to affect Milosevic’s

risk calculus, VJ units were reported in mid-May to be digging in
along likely attack routes from Macedonia and Albania and fortifying
the border, in a distinct shift in effort from expelling ethnic Albanians
to preparing for a possible showdown with NATO on the ground. In
______________
12Dana Priest, “A Decisive Battle That Never Was,” Washington Post, September 19,
1999.
13 Patrick Wintour and Peter Beaumont, “Revealed: The Secret Plan to Invade
Kosovo,” London Sunday Observer, July 18, 1999.
14Peter Beaumont and Patrick Wintour, “Leaks in NATO—and Plan Bravo Minus,”
London Sunday Observer, July 18, 1999.
particular, VJ troops were observed laying mines and attempting to

block potential ground attack routes from Skopje and Kumanovo in
Macedonia, in a pattern of activity suggesting that the allied bomb-
ing effort had not yet come close to breaking their cohesion and
fighting spirit. 15
Moreover, earlier on the same day that Milosevic eventually capitu-

lated, President Clinton held a widely publicized meeting with his
service chiefs for the express purpose of airing options for land force
employment in case NATO decided it had no choice but to approve a
ground invasion.16 That was his first meeting with all four chiefs at
any time during the course of Operation Allied Force. Immediately
after the meeting, which left the issue unresolved, Clinton was said to
have been planning to inform the chiefs that he was now ready to
sign on to a ground invasion should developments leave no alterna-
tive.17 In what he later described as “a pretty depressing memo” to
the president, Berger wrote that “we basically should go ahead with
what Clark had proposed if the [Ahtisaari-Chernomyrdin] mission
failed.” In that memo, Berger listed three options. The first, to arm
the Kosovars, would create a multitude of undesirable downstream
consequences that would persist for years and thus was ruled out as
a nonstarter. The second, to wait until spring, was equally unaccept-
able because it would oblige NATO to supply and protect the Kosovar
refugees in Albania throughout the winter. That left only the third
option, a massive ground invasion by 175,000 NATO troops, some
100,000 of whom would be American. 18 Taken together, these de-
velopments made for a compelling pattern of evidence suggesting
that both Washington and its chief NATO allies had crossed the Ru-
bicon when it came to facing up to the land-invasion issue, and that
they had become determined by the end of May to commit to a
______________
15Michael R. Gordon, “NATO Says Serbs, Fearing Land War, Dig In on Border,” New
York Times, May 19, 1999.
16Jane Perlez, “Clinton and the Joint Chiefs to Discuss Ground Invasion,” New York
17For details, see Steven Erlanger, “NATO Was Closer to Ground War in Kosovo Than
Is Widely Realized,” New York Times, November 7, 1999.
18McManus, “Clinton’s Massive Ground Invasion That Almost Was.” NATO com-
manders were asking for three months to assemble the invasion force.
forced entry on the ground if the bombing did not produce an ac-
ceptable settlement soon.
Some, however, have made more of this sequence of events than the
evidence warrants. In the early wake of the successful conclusion of
Operation Allied Force, revisionist claims began emanating from
some quarters suggesting that the air effort had been totally ineffec-
tive and that, in the end, it had been Milosevic’s fear of a NATO
ground invasion that induced him to capitulate.19 Clark himself, in
his memoirs, indicated his belief that by mid-May, NATO “had gone
about as far as possible with the air strikes” and that in the end, it
had been the Apache deployment and the prospect of a NATO
ground intervention that, “in particular, pushed Milosevic to con-
cede.”20 That notwithstanding the all-but-conclusive evidence Clark
presented elsewhere throughout his book that NATO’s top political
leaders were nowhere near having settled on a definitive invasion
plan—let alone decided to proceed with such a plan should the
bombing prove unavailing.21 Even viewed in the most favorable light
conceivable, such far-reaching claims on behalf of the implied
ground threat defy believability because any NATO land invasion,
however possible it may eventually have been, would have taken
months, at a minimum, to prepare for and successfully mount.
In contrast, Milosevic was living with the daily reality of an increas-

ingly brutal air war that showed no sign of abating. Although Clark’s
______________
19A recent example of this countercontention dismissed the claims of unspecified “air
power enthusiasts” and posited instead that “the decision to commit ground forces [a
decision which, in fact, had not been made at the time of Milosevic’s capitulation] was
critical to NATO’s success.” Brigadier General Huba Wass de Czege, USA (Ret.) and
Lieutenant Colonel Antulio J. Echevarria II, USA, “Precision Decisions: To Build a
Balanced Force, the QDR Might Consider These Four Propositions,” Armed Forces
Journal International, October 2000, p. 54.
20General Wesley K. Clark, Waging Modern War: Bosnia, Kosovo, and the Future of
Combat, New York, Public Affairs, 2001, pp. 305, 425, emphasis added.
21The most compelling of such evidence cited by Clark was the May 28 statement by
Secretary of Defense Cohen, made less than a week before Milosevic capitulated, that
“there is no consensus for a ground force. And until there is a consensus, we should
not undertake any action for which we could not measure up in the way of perfor-
mance. . . . And so, there is a very serious question in terms of trying to push for a
consensus that you really diffuse or in any way diminish the commitment to the air
campaign. The one thing we have to continue is to make sure we have the allies
consolidated in strong support of the air campaign. They are. And they are in favor of
its intensification. So that’s where we intend to put the emphasis.” Ibid, p. 332.
effort to find and attack dispersed and hidden VJ forces in Kosovo

was consuming the preponderance of shooter sorties while accom-
plishing little by way of tangible results, more and more infrastruc-
ture targets were also being approved and struck every day.22 In a
revealing admission of what was uppermost among his concerns on
the day he elected to settle, Milosevic asked Chernomyrdin directly
on June 3 in response to NATO’s ultimatum: “Is this what I have to
do to get the bombing stopped?” Chernomyrdin replied in the affir-
mative, with Ahtisaari adding: “This is the best you can get. It’s only
going to get worse for you.” To which Milosevic responded: “Clearly
I accept this position.”23
There is no question that by the end of May, NATO had yielded to the
inevitable and embraced in principle the need for a ground invasion
should the bombing continue to prove indecisive. There also is every
reason to believe that awareness of that change in NATO’s position
on Milosevic’s part figured importantly in his eventual decision to
capitulate. There is no basis, however, for concluding that the mere
threat of a land invasion somehow overshadowed the continuing,
here-and-now reality of NATO’s air attacks as the preeminent con-
sideration accounting for that decision. There also is little benefit to
be gained from the misguided efforts by air and land power partisans
alike to argue the relative impact of the air attacks and ground threat
in simplistic either-or terms. It detracts not in the least from the air
war’s signal accomplishments to concede that developments on the
land-invasion front almost surely were part of the chemistry of Milo-
sevic’s concession decision. Although any impending ground inter-
vention was months away at best, there is no question that both the
Clinton administration and the principal NATO allies had made up
______________
22However, by dispersing their assets and selectively emitting with their radars, Serb
IADS operators forced NATO aircrews to remain wary to the very end and denied them
the freedom to operate at will in hostile airspace. Although the Serbs’ repeated at-
tempts to bring down NATO aircraft frequently came in the form of ineffective ballistic
launches, the launches were amply disconcerting to allied pilots, who were forced to
threat-react—often aggressively—to ensure their own safety. Many guided shots in
accordance with IADS doctrine were also fired against attacking allied aircraft, requir-
ing even more aggressive and hair-raising countertactics by the targeted aircraft. A
first-hand account of one such episode is reported in Dave Moniz, “Eye-to-Eye with a
New Kind of War,” Christian Science Monitor, March 23, 2000.
23Quoted in Tyler Marshall and Richard Boudreaux, “Crisis in Yugoslavia: How an
Uneasy Alliance Prevailed,” Los Angeles Times, June 6, 1999.
their minds on the need to do something along those lines should the
air war continue to prove unavailing. In light of that, as two RAND
colleagues have suggested, “in assessing NATO air attacks on Serbia,
analysts should focus not on the role air power played instead of a
ground invasion . . . but on the role it played in combination with the
possibility of one.”24
MILOSEVIC’S PROBABLE DECISION CALCULUS

To better understand the interaction of influences that most likely
persuaded Milosevic to concede, it may be instructive to view Allied
Force as it unfolded not through our own frame of reference, but
rather through Serbian eyes. Those who planned and ran the air op-
eration understandably tended to fixate on such negative aspects as
target-list restrictions and what many considered to be excessive
fretfulness on the part of the alliance’s political leaders over the pos-
sibility of causing collateral damage. For them, the air war’s domi-
nant hallmarks were such sources of daily frustration as repeated
delays in the target approval process and the consequent inefficiency
of the overall effort. Naturally, in their view, the performance of air
power in Operation Allied Force left a great deal to be desired.
Yet to those on the operation’s receiving end far removed from such
concerns, it must have seemed, certainly by the end of the second
month, as though NATO was prepared to keep escalating and to
continue bombing indefinitely. From Milosevic’s viewpoint, new
targets were being attacked with mounting regularity after the NATO
summit of April 23–25, and ever more infrastructure targets were be-
ing hit with seemingly no end in sight. Moreover, one might surmise
that even the inadvertent Chinese embassy bombing played an indi-
rect part in inducing Milosevic to capitulate. Whatever U.S. and
NATO officials said about that incident for the public record, Milo-
sevic may have thought that the bombing had been intentional and
that it presaged both a lifting of NATO’s target limitations and worse
damage yet to come. As if to affirm that fear after the fact, USAFE’s
commander, General John Jumper, later disclosed that with the in-
creased number of strike aircraft that had become available in the-
______________
24Daniel L. Byman and Matthew C. Waxman, “Kosovo and the Great Air Power De-
bate,” International Security, Spring 2000, p. 15.
ater by late May, the operation’s intent was to employ FACs and be-
gin attacking kill boxes all throughout Serbia, not just in Kosovo, and
to go at will after tunnels, bridges, storage areas, and other military
targets of interest.25
The almost universal belief among air warfare professionals that a

more aggressive effort starting on opening night, in consonance with
a more doctrinally pristine strategy, would have yielded the same re-
sult more quickly may have been correct as far as it went, but that
conviction was based solely on faith in the intrinsic power of the air
weapon, not on any evidence directly related to the case at hand.
The only way a more intensive and resolute air campaign would have
caused Milosevic to fold substantially sooner than he did would have
been for the air war’s effects to persuade him that much earlier that
his strategy had no chance of succeeding.
In fact, as RAND colleague Stephen Hosmer has argued, Milosevic’s

decision to capitulate hinged on developments that necessarily took
time to unfold and mature.26 To begin with, the Serb dictator, just
like NATO, pursued a concrete, if also flawed, strategy from the very
start. He knew that the terms levied by the United States at Ram-
bouillet, if implemented, would have replaced Serb dominance over
Kosovo with a NATO military presence that claimed rights of access
to all of Yugoslavia. They also would have raised the distinct pos-
sibility that Kosovo’s future would be decided by a NATO-enforced
referendum, an event which could only have resulted in a loss for
Serbia.27 Those two threatened outcomes, along with additional
downside consequences, would have put at risk not only Serbian
control over Kosovo, but also the foundations of Milosevic’s personal
rule, and hence his political—and perhaps even physical—survival.
______________
25General John Jumper, USAF, “Oral Histories Accomplished in Conjunction with
Operation Allied Force/Noble Anvil.”
26Stephen T. Hosmer, The Conflict over Kosovo: Why Milosevic Decided to Settle When
He Did, Santa Monica, California, RAND, MR-1351-AF, 2001.
27The latter of these two concerns was more an issue for Milosevic than the former.
Had he been seriously worried about a NATO presence that might actually encroach
into Serbia, as opposed to just taking effective control of Kosovo (his real fear),
he would have sought to head off that possibility at Rambouillet. He never did. I
am grateful to Ivo Daalder of the Brookings Institution for bringing this point to my
attention.
In addition, Milosevic probably convinced himself that if he hun-

kered down and stoically endured the bombing, he could undermine
NATO’s persistence and cohesion by ensuring the eventual occur-
rence of noncombatant civilian fatalities and extracting the fullest
propaganda value from collateral-damage incidents. Indeed, he
most likely balked at Rambouillet in full expectation that he would be
bombed by NATO, yet only symbolically and for a token period of
time, convinced that NATO would lack the stomach to continue
bombing for very long. On this point, Stojan Cerovic, a Serb journal-
ist working in Washington, suggested that Milosevic at first saw no
danger to himself from the bombing and operated on the assump-
tion that other nations would become so incensed over NATO’s per-
ceived attempts at hegemony that they would rally behind the Serb
cause.28 No doubt expecting nothing more than a replay of the inef-
fectual pinprick attacks that had been carried out by U.S. forces
against Iraq since the preceding December, he evidently calculated
that he could easily wait out any punitive air strikes that NATO might
bring itself to carry out.
Where Milosevic blundered even more grievously than did NATO (in
the latter’s faulty assumption that just a few days of bombing would
suffice) was in unleashing the full brunt of his ethnic cleansing cam-
paign almost immediately after Allied Force began. No doubt he cal-
culated that Operation Horseshoe would quickly empty Kosovo of its
ethnic Albanian populace and thus enable him to move directly
against the KLA, eliminate it as a continued factor affecting any ulti-
mate political outcome, and, along the way, solve his ethnic problem
in Kosovo with a fait accompli. Alternatively, or perhaps in addition,
he may also have been trying to signal his own determination to
NATO, although there is no “smoking-gun” evidence to this effect.
After all, the main lesson he likely drew from Deliberate Force in 1995
was that he gave up the fight just a few days too early. Most assess-
ments of Deliberate Force include arguments that NATO was ap-
proaching the end of its rope politically and militarily because of a
______________
28 Justin Brown, “Why U.S. Bombs Failed to Topple Milosevic,” Christian Science
Monitor, March 24, 2000.
lack of additional approved targets. 29 In light of that perception,

Milosevic, in addition to working on his Kosovar Albanian problem,
may simply have been trying to tell everyone that this time it would
not be so easy. Whatever the case, his depredations instead merely
galvanized NATO’s resolve and ensured that the allies would con-
tinue bombing until their objectives were met. By throwing down a
gauntlet to NATO and, in effect, challenging it to see who could hold
out longer, Milosevic forced NATO to recognize that its own credibil-
ity and existence as an alliance were now on the line.30
There is no way of knowing for sure from the evidence currently

available why June 3 was the date on which Milosevic finally elected
to give in. There is a strong presumptive case to be made, however,
that by the end of May, he had come to realize that any remaining
countercoercive leverage he had over NATO was almost nonexistent.
As Hosmer concluded, once the Serb dictator became convinced that
future attacks would be unconstrained, a settlement at the earliest
possible moment became not just an option but an imperative.
Continued bombing during the negotiations over implementation of
the agreement, moreover, closed the door to any possibility of his
backsliding. Milosevic further had every reason to assume by that
time that any terms of a settlement agreement would never look
better, and that the time was propitious for a loss-cutting move while
he could retain at least the polite fiction of having extracted conces-
sions from NATO.
As for disincentives against holding out any longer, Milosevic also

had every reason to believe that continued resistance on his part
would only lead to continued, and quite probably escalated, bomb-
ing. Even in the absence of an imminent NATO ground assault, he
knew that the air war could have continued for many more weeks,
even indefinitely. With the possibility that electrical power and water
supplies to Belgrade might be cut off at any time, the approach of
______________
29See, in particular, Colonel Robert Owen, USAF, ed., Deliberate Force: A Case Study
in Effective Air Campaigning, Maxwell AFB, Alabama, Air University Press, January
2000, pp. 455–522.
30Stephen Hosmer has pointed out that the ethnic cleansing hardened NATO’s re-
solve in another way as well: Only a NATO military presence in Kosovo would have
convinced the refugees to go back to their homes, and no outcome short of the latter
would have been acceptable to NATO.
winter offered the prospect of making daily life horrendously difficult

for Serbia’s leaders and rank and file alike. Worse yet, the mere
thought of a NATO land invasion occurring at some indeterminate
future point had the most ominous implications, in that it could have
meant Serbia’s loss of Kosovo for good, posing the direst threat to
Milosevic’s survival. In light of those mutually reinforcing facts, he
evidently convinced himself that although his own continued liveli-
hood required his capitulation, he could convert his tactical defeat
into a long-term loss for NATO by swallowing his temporary setback
in Kosovo while remaining in power to fight another day.
In sum, although it did not achieve a military victory over Belgrade in

the classic sense, NATO unquestionably prevailed over Milosevic in a
high-stakes contest of wills. Diplomacy and coercive bombing to-
gether convinced the Serb dictator that he had failed to split NATO
and that Russia would not act to stop the air war. At the same time,
they allowed him enough maneuver room to maintain at least a fig
leaf of a claim to credibility in the eyes of his compatriots that he had
not yielded to NATO on all fundamentals. As Barry Posen concluded,
“all of the principal wedges into NATO’s cohesion had been tested.
Further testing would prove very expensive in terms of damage to
Serbia’s infrastructure and economy.”31
In the end, however inefficient the air war may have been because of
its need to honor U.S. and NATO domestic political realities, the
manner in which it was conducted (avoiding friendly fatalities and
minimizing noncombatant enemy casualties) nevertheless effec-
tively countered and ultimately neutralized Milosevic’s strategy by
keeping NATO’s cohesion intact to the very end. In response, the
Serb dictator most likely opted to accept NATO’s demands simply
out of a rational calculation that he had nothing to gain and much to
risk by holding out any longer. Indeed, as the endgame neared, one
can imagine how he may even have begun to harbor dark visions of
being gunned down in the street, in the grim manner of the Ceauces-
______________
31Barry R. Posen, “The War for Kosovo: Serbia’s Political-Military Strategy,” Interna-
tional Security, Spring 2000, p. 75. One can, however, question Posen’s subsequent
suggestion that Milosevic achieved “some political success” by holding out as long as
he did, considering that he lost control of Kosovo, suffered heavy damage to his infra-
structure and economy, and ultimately was defeated in a fair election, arrested, and
jailed for having committed crimes against the state.
cus after their control over Romania collapsed in 1991. Said a source
close to the Yugoslav government: “I can’t pinpoint an exact
moment when Milosevic finally listened, but there was tremendous
pressure from all sides; the West, his inner circle, and his wife. It was
building up, and eventually he just let go.”32
THE DETERMINING ROLE OF THE AIR WAR

To repeat a point stressed at the beginning of this chapter, it would
be reductionist to a fault to conclude that Milosevic was bombed into
submission by air attacks to the exclusion of any other contributing
factors. However, the bombing did create political conditions in Bel-
grade that enabled Milosevic to negotiate. 33 Insofar as the bombing
may have been insufficient to produce his capitulation in and of it-
self, it bears underscoring that those conditions were all indirect ef-
fects of the air war. Had it not been for Allied Force and its direct
effects, the additional stimuli would never have materialized. As
General Clark later remarked, “the indispensable condition for all
other factors was the success of the air campaign itself.”34
From the Yugoslav perspective, there must have been a nagging

sense of the inexorability of NATO’s eventual victory as the air war
neared the end of its second month. The truculent early defiance
that was so studiously expressed by Belgrade’s citizens before the
war began affecting them personally soon turned into sullen resig-
nation under the mounting duress caused by the bombing of infra-
structure targets. For a time, the half-hearted bombing during the
first month actually seemed to rally public determination to with-
stand the offensive and to increase public support for the widely un-
popular Milosevic. However, the spontaneous street celebrations
that erupted immediately after the cease-fire suggested that the Yu-
goslav rank and file had begun to doubt Milosevic’s stewardship in
having led the country into an unwinnable contest of wills against
the world’s most powerful alliance. Possibly reflecting mounting
______________
32“NATO’s Game of Chicken,” Newsweek, July 26, 1999, p. 59.
33For detailed amplification on this point, see Hosmer, The Conflict over Kosovo: Why
Milosevic Decided to Settle When He Did.
34 Quoted in John T. Correll, “Lessons Drawn and Quartered,” Air Force Magazine,
December 1999, p. 2.
popular weariness of the bombing, Deputy Prime Minister Vuk

Draskovic declared as early as April 25 that “Yugoslavia should rec-
ognize that it cannot defeat NATO and that it must face the reality of
a world standing against Yugoslavia.”35
The precise and measured nature of the attacks that were being con-
ducted against leadership and infrastructure targets in the heart of
the Yugoslav capital on a daily basis only became fully apparent to
outside observers after they had a chance to inspect the results up
close. As one American reporter who visited Belgrade after the war
remarked tellingly: “Like ice-pick punctures in the neck, the chilling
quality of the strikes was not their size but their placement. We
stopped at an intersection in the heart of the city. At each corner of
the intersection, but only at each corner, there were ruins. The Ser-
bian government center, the foreign ministry and two defense min-
istry buildings had been reduced to rubble or were fire-gutted shells.
The precision of the destruction suggested a war with an invisible,
all-seeing enemy and a city helpless to protect itself.”36
In what may have been read by Milosevic as an ominous indicator

that the bombing was coming ever closer to the most senior national
leadership, General Ljubisa Velichkovic, the former air force chief of
staff, was killed in an air attack on Day 70 while visiting VJ troops in
the field. Velichkovic, who had been removed from office by Milo-
sevic the previous year as a part of a purge of the military leadership
and been given the honorific title of deputy chief of staff, was identi-
fied as the highest-ranking casualty since Operation Allied Force be-
gan.37 It is entirely possible that Milosevic had come to fear by that
point that a similar fate could befall him at any moment.
Viewed in hindsight, the bombing seems to have had two outcome-

determining effects. First, it eventually persuaded Milosevic that
______________
35Quoted in Robert Hewson, “Allied Force, Part II: Overwhelming Air Power,” World
Air Power Journal, Winter 1999/2000, p. 97. Three days later, Draskovic was fired by
Prime Minister Momir Bulatovic for having made that statement.
36Blaine Harden, “The Milosevic Generation,” New York Times Magazine, August 29,
1999, p. 34.
37“Sacked Yugoslav Air Chief Killed,” London Times, June 2, 1999. See also William
Drozdiak and Steven Mufson, “NATO Sending Tough Terms to Belgrade,” Washington
Post, June 2, 1999.
NATO not only would not relent, but also was determined to prevail
and had both the technical and political wherewithal to do so. Sec-
ond, given the incapacity of the Serb IADS to shoot down significant
numbers of allied aircraft, it further convinced him that his own de-
feat sooner or later was inevitable. Although its resolve was slow in
coming, NATO finally showed that it would not be moved by the
public outcry over collateral damage and could sustain the bombing
indefinitely, at a negligible cost in terms of friendly losses. As with
Iraq’s forces during Operation Desert Storm, the VJ’s leaders, no less
than Milosevic, must have found NATO’s ability to inflict unrelenting
damage on their country with virtual impunity to be profoundly de-
moralizing. Before June 3, the commander of the VJ’s 3rd Army in
Kosovo, General Nebojsa Pavkovic, had argued that his forces re-
mained more or less intact and that they could defend Serbia if put to
the test. After Ahtisaari and Chernomyrdin delivered NATO’s ultima-
tum on June 2 and a cease-fire was agreed to, however, he reportedly
declared to a group of disconcerted VJ reservists that Serbia’s leaders
had been put on notice by the Russians that if NATO’s terms were
rejected, “every city in Serbia would be razed to the ground. The
bridges in Belgrade would be destroyed. The crops would all be
burned. Everyone would die.”38
True enough, thanks to the improved flexible targeting procedures

(that is, procedures for responding promptly to mobile or pop-up
targets that had been detected by allied sensors) that had been im-
plemented by late April (see Chapters Six and Seven) and the clearer
weather that had begun to develop the following month, NATO’s
ability to get at dispersed and hidden enemy forces in Kosovo im-
proved perceptibly during the air war’s final week. In all likelihood,
however, NATO broke Milosevic’s will and that of his political sup-
porters primarily because it had convincingly shown that it could
also destroy such key infrastructure targets as hardened bunkers,
bridges, electrical power stations, and other targets directly tied to
Yugoslav society and the regime’s control over it. By all indications,
those attacks played the central role in bringing Milosevic to accept
NATO’s demands and created the political conditions in Serbia that
______________
38 Quoted in Chris Hedges, “Angry Serbs Hear a New Explanation: It’s All Russia’s
Fault,” New York Times, July 16, 1999.
allowed Milosevic to abandon Belgrade’s physical presence in

Kosovo in exchange for a cessation of the bombing.
As one may recall, manipulation of the Kosovo issue and Serbia’s

strong emotional attachment to the province had figured promi-
nently in Milosevic’s rise to power and in his continued hold on it
since 1989. For that reason, acceding to NATO’s demands as ex-
pressed in the proposed Rambouillet accords would, in all likelihood,
have meant political suicide for him. By June 1999, the opposite had
become true: Milosevic’s continued survival seemed to depend on
finding a way to stop the bombing and to extricate himself gracefully
from his growing predicament. Although Ahtisaari and Cherno-
myrdin provided him with the ready pretext that he needed, it was
the air war’s steadily increasing encroachment on Serbia’s core eq-
uities that most likely prompted the decisive shift in his political cal-
culus, as perhaps best attested by his own plaintive question to
Chernomyrdin on June 2 cited earlier.39 In contrast, by Clark’s own
admission after the cease-fire, the attempted attacks against dis-
persed and hidden VJ forces in Kosovo caused the latter little signifi-
cant pain or inconvenience. That suggests, by elimination, that
whatever one may believe was Milosevic’s most critical vulnerability,
the bombing of Clark’s target priorities in the KEZ was not what
mainly swung his decision to capitulate. 40
On this still-contentious issue, defense analyst William Arkin, who

led a private bomb damage assessment mission for Human Rights
Watch for three weeks in August 1999 and who visited more than 250
targeted sites in the process, perhaps offered the most helpful and
______________
39Indeed, from a low of fewer than 100 daily strike sorties flown during the air war’s
fifth night, the bombing effort intensified steadily and uninterruptedly to almost three
times that number by the eve of Milosevic’s capitulation on June 3. Briefing by the
chairman of the Joint Chiefs of Staff, June 10, 1999, cited in Major General Eitan Ben-
Eliahu, commander, Israeli Air Force, “Air Power in the 21st Century: The Impact of
Precision Weapons,” Military Technology, April 2000, p. 40.
40It bears acknowledging here, however, that only the authoritative report of NATO’s
intent to proceed with an eventual ground invasion, should the bombing alone fail to
dislodge Milosevic, finally convinced Moscow to play its constructive role in June
1999. Russia’s deploying of Chernomyrdin helped negotiate an international military
presence in Kosovo, thus warding off a NATO-only presence and preserving at least
some Russian influence in the Balkans. On this point, see the informed comment of-
fered by former Russian foreign ministry Balkan official Oleg Levitin, “Inside
Moscow’s Kosovo Muddle,” Survival, Spring 2000, p. 138.
incontestable perspective when he observed: “It was not what we

bombed, but that we bombed. The coalition didn’t crumble, the
Russians didn’t bail Belgrade out, China was unable to affect the war.
At some point it was clear to Milosevic that he wasn’t going to be able
to wait out the bombing, that NATO wasn’t going to go away, and
that progressively Serbia was being destroyed, he chose to get the
best negotiated settlement he could. To say it was this or that target
that was important to Milosevic is just to engage in mirror-image
speculation.”41
______________
41William Arkin, “Yugoslavia Trip Report,” September 8, 1999. In a similar vein, Karl
Mueller suggested that “while it was not clear how NATO was going to win, it certainly
would continue the effort until it managed to do so. From this perspective, it was
not what NATO was bombing that mattered, but the fact that it was continuing to
bomb. . . .” Karl Mueller, “Deus ex Machina? Coercive Air Power in Bosnia and
Kosovo,” unpublished paper, School of Advanced Air Power Studies, Maxwell AFB,
Alabama, November 7, 1999, p. 10.
Chapter Five
ACCOMPLISHMENTS OF THE AIR WAR
A number of “firsts” were recorded during NATO’s air war for

Kosovo. To begin with, Operation Allied Force was the first air war in
which all three currently deployed U.S. Air Force heavy bomber types
saw combat use. Those bombers constituted a major part of the
overall strike force. Of some 700 U.S. combat aircraft committed to
the operation altogether, a mere 21 heavy bombers (10 B-52s, 5 B-1s,
and 6 B-2s) delivered 11,000 out of the more than 23,000 U.S. air-to-
ground munitions that were expended over the operation’s 78-day
course.1
There also was an unprecedented use of precision-guided munitions

in the air war. In Desert Storm, only 10 percent of the participating
U.S. strike aircraft were PGM-capable. That number rose to 69 per-
cent in Operation Deliberate Force and shot up to 90 percent in Al-
lied Force.2 Thanks to the heavy use of PGMs in the interest of both
operational efficiency and avoiding unintended collateral damage, a
full three-quarters of the more than 400 fixed targets attacked in
Serbia were assessed as having sustained moderate to severe dam-
age.3 Some 64 percent of the 9,815 aim points altogether were hit
by PGMs, for a total hit rate of 58 percent.4 Figure 5.1 shows the
______________
1David Atkinson, “B-2s Demonstrated Combat Efficiency over Kosovo,” Defense Daily,
July 1, 1999, p. 1.
2Comments on an earlier draft by Hq USAFE/SA, April 6, 2001.
3Kosovo: Lessons from the Crisis, Report to Parliament by the Secretary of State for De-
fense, The Stationery Office, London, England, June 2000, p. 36.
4“AWOS Fact Sheet,” Hq USAFE/SA, December 17, 1999.
87
RAND MR1365-5.1
Total: 23,315
PGM
(6,728)
29%
Non-PGM 71%
(16,587)
Figure 5.1—U.S. Precision and Nonprecision Munitions Expended
proportion of precision munitions and nonprecision munitions de-

livered by U.S. combat aircraft over the 78-day course of the bomb-
ing effort. At nearly a third of the total number of ground-attack
munitions expended altogether, PGM use in Allied Force greatly
overshadowed that in Operation Desert Storm nearly a decade ear-
lier. In that conflict, the proportion of PGMs delivered by U.S. forces
compared to nonprecision munitions was less than 10 percent.
In addition, more than in any previous U.S.-led air operation, UAVs

were used in Allied Force for combat support, most notably for locat-
ing VJ troops dispersed in the KEZ.5 In yet another precedent, the
______________
5The qualification “U.S.-led” is appropriate here, considering that the Israeli Air Force
has made regular and highly effective use of UAVs over southern Lebanon for nearly
two decades, going back to the Beka’a valley air campaign of 1982.
Accomplishments of the Air War 89
USAF’s air expeditionary force (AEF) concept was first successfully

exercised in a full-up combat setting, with expeditionary fighter
squadrons deploying to Aviano Air Base, Italy, from the continental
United States and from U.S. bases in Europe and folding into the an-
chor 31st Air Expeditionary Wing stationed there, which, at its peak,
operated a record 175 combat aircraft.6 Relatedly, the assignment of
tactical control of 12 C-17s directly to USAFE roundly validated that
aircraft’s “direct delivery” status and reflected a major step forward
in the employment of air mobility forces as global assets. Finally, as
the discussion below will sketch out in more detail, Operation Allied
Force saw the most extensive use of space systems in combat to date,
with more than 50 U.S. and European satellites directly involved in
support of USEUCOM and NATO intelligence, coordination, and at-
tack activities.
THE COMBAT DEBUT OF THE B-2

Of major note, Allied Force finally saw the long-awaited combat de-
but of the USAF’s B-2 stealth bomber, which was the first manned
aircraft to penetrate Serb air defenses the first night. 7 As the final
countdown drew near, expectations ran high throughout the Air
Force that the regional combatant commanders in chief around the
world, who had long resisted the B-2’s use in earlier air power appli-
cations because of their distrust of unproven systems, would finally
be won over by a record of unblemished accomplishment by the air-
craft over Serbia and Kosovo. Those expectations were more than
vindicated. Of 19 B-2s all told that had been delivered to the air-
craft’s parent 509th Bomb Wing at Whiteman AFB, Missouri, only 9
were available to USEUCOM for combat operations, with the other
10 undergoing avionics upgrades to the aircraft’s definitive Block 30
status. 8 Nevertheless, to the surprise of many, the B-2 turned out to
be the most consistently effective performer of the entire air war.
According to the 509th commander at the time, Brigadier General
______________
6The wing had most of the essential support assets on hand, so deploying squadrons
did not need to bring much by way of logistics overhead.
7Dale Eisman, “Over Balkans, It’s Beauty vs. the Beast,” Norfolk Virginian-Pilot, April
26, 1999.
8Vince Crawley, “B-2s See Combat over Yugoslavia,” Defense Week, March 29, 1999,
p. 6.
Leroy Barnidge, B-2 operations demonstrated a 96-percent weapons

effectiveness rate.
Since only six of the nine available aircraft were actually used on
combat missions, the average turn time per aircraft was two days.9
There was never a shortage of capability to meet USEUCOM’s target-
ing needs, however. Some B-2s were turned in the time it took to
refuel them. The only reported case of a B-2 component having
failed during a combat mission was a malfunction of a rotary bomb
launcher, which was promptly repaired upon the aircraft’s return to
base.10 The chief maintenance drivers were said to have been the
aircraft’s low-observable treatment, its flight control system, its
synthetic-aperture radar, and engine accessory drives.
Each B-2 flew nonstop to its targets in its final Block 30 configuration
directly from Whiteman on 28- to 32-hour round-trip missions, de-
livering up to 16 global positioning system (GPS)-guided GBU-31
joint direct-attack munitions (JDAMs) from 40,000 ft, usually through
cloud cover, against enemy targets including hardened command
bunkers and air defense facilities. Those missions typically entailed
15-hour legs out and back, with two inflight refuelings per leg. Two
aircraft were launched on 15 nights and just a single aircraft on 19
nights. The aircrews quickly adjusted to these unprecedentedly long
missions and coped with them adequately. They also quickly
adapted to the demands of real-time targeting changes en route.
Although the USAF bomber community, by virtue of its traditional
nuclear focus, had long been predisposed to do things in a carefully
preplanned way, USAFE’s commander, General John Jumper, trav-
eled to Whiteman and personally talked to B-2 aircrews about the
need for rapid adaptability. After just a few hours of intense opera-
tor-to-operator brainstorming, any residual doubts some B-2 pilots
may have harbored regarding the merits of replacing traditional
cold-war practices with real-time improvisation as needed to meet
current demands were put to rest. The first time the ensuing air ef-
______________
9Of the nine available B-2s at Whiteman, one was kept aside for training, one was un-
dergoing final upgrades to Block 30 status, and one was in extensive maintenance.
“Missouri-to-Kosovo Flights for B-2 Not a Concern to Wing Commander,” Inside the
Air Force, July 2, 1999, p. 12.
10“B-2 Performed Better in Kosovo Than USAF Expected,” Inside the Pentagon, July 8,
1999, p. 11.
fort attempted to apply what came to be called “flex” (for flexible)

targeting against enemy assets that had been detected and identified
only on short notice, the B-2s took out two SA-3 sites that had been
assigned to them only a few hours prior to their planned arrival over
target.11
In all, 49 B-2 combat sorties were launched out of Whiteman, of

which 45 made it to target and were cleared to drop munitions. Al-
though that was less than half a percent of the 9,500 strike sorties
flown in Allied Force altogether, the B-2 dropped 11 percent (some
700) of the bombs delivered against fixed targets in Serbia and
Kosovo. It also dropped a full third of all precision munitions ex-
pended during the air effort.12 In addition to its normal load of
JDAMs, the B-2 was also configured to carry the GPS-guided GBU-37
for special missions against deeply buried or superhardened tar-
gets.13 A total of 652 JDAMs and 4 GBU-37s were dropped, with
more than 80 percent of the B-2’s assigned targets being hit on a
single pass. 14 In a major improvement in the combat leverage and
versatility of the American air weapon, the aircraft proved itself ca-
pable of operating effectively above weather that grounded all other
allied combat aircraft. It also consistently achieved up to 16 separate
target hits per sortie.
It bears emphasizing here that the B-2 did not merely drop weapons
preprogrammed to home in on assigned coordinates, but used its
onboard synthetic aperture radar (SAR) to take two successive
images of the target during its initial approach. By so doing, the B-2
was able to eliminate the largest target error source in the JDAM,
namely, the error in the exact location of the aim point in GPS space.
As a result, the B-2’s average miss distance with JDAM was less than
half the 13 meters stipulated for unassisted JDAMs.15
______________
11“Jumper on Air Power,” Air Force Magazine, July 2000, p. 43.
12Paul Richter, “B-2 Drops Its Bad PR in Air War,” Los Angeles Times, July 8, 1999.
13 Adam Hebert, “Air Force Follows Roadmap in Employment of Bombers Against
Serbia,” Inside the Air Force, April 2, 1999, p. 2.
14Barry D. Watts, “The EA-6B, E-8C, and B-2 in Operation Allied Force,” Northrop
Grumman Analysis Center briefing, Rosslyn, Virginia, May 8, 2000.
15Barry D. Watts, The Military Use of Space: A Diagnostic Assessment, Washington,
D.C., Center for Strategic and Budgetary Assessments, February 2001, p. 42.
On most nights, penetrating B-2s received standoff jamming support

from Navy or Marine EA-6Bs, as well as SEAD support from orbiting
F-16CJs standing by as needed as a precautionary measure. On at
least one occasion, however, B-2 strikes occurred without any off-
board jamming support. Thanks to the aircraft’s third-generation
stealth properties, it did not require such support to ensure its sur-
vivability, and EA-6B jamming for both the B-2 and the F-117 was
said to have been “indirect.” Supporting EA-6B and F-16CJ pilots
were provided with time blocks and rough areas within which the
stealthy aircraft would be concurrently operating, but not the exact
routing of those aircraft. In the absence of those mission specifics,
they relied on time and space deconfliction to maintain safe separa-
tion.16 Because of their low observability and the persistence of
overlapping and unlocated enemy SAM defenses, only the B-2s and
F-117s were committed against targets in downtown Belgrade for the
first 58 days of the operation.17
Since every B-2 mission, save one or two, benefited from dedicated
offboard electronic countermeasures (ECM) support and was flown
against less than top-of-the-line enemy defenses, it remains unclear
as to what extent the aircraft’s stealth properties were truly tested in
modern combat. However, by all accounts the aircraft was never
tracked by enemy radar, let alone shot at by enemy SAMs. Unlike all
other aircraft that flew combat missions in Allied Force, the B-2 op-
erated autonomously. It simply checked in with the ABCCC as it ap-
proached the target area, received a go/no-go code, and pressed
ahead to its assigned targets in radio silence. If a target change was
required en route, the Combined Air Operations Center (CAOC)
could pass essential information to ingressing B-2 aircrews as much
as an hour and 45 minutes before the aircraft’s scheduled time on
target (TOT). That ability to select new targets while airborne en-
abled the aircraft to take out some enemy SA-3s and their radars
shortly after they were located and identified by allied sensors.18 The
______________
16Watts,” The EA-6B, E8C, and B-2 in Operation Allied Force.”
17Colonel Tony Imondi, 509th Operations Group commander, quoted in Bill Sweet-
man, “B-2 Is Maturing into a Fine Spirit,” Jane’s International Defense Review, May
2000.
18Brigadier General Randy Gelwix, USAF, “Oral Histories Accomplished in Conjunc-
tion with Operation Allied Force/Noble Anvil.”
B-2’s onboard GPS-aided targeting system (GATS) and SAR also

allowed the aircraft to find, identify, and successfully attack impre-
cisely located targets.
As the air war unfolded, former Secretary of the Air Force Donald
Rice observed that the B-2, although one of the most controversial
weapons in the U.S. inventory, was “proving to be the nation’s single
most cost-effective attack aircraft.” 19 Rice further pointed out that
the much-derided stealth treatments on the aircraft had proven
themselves durable and reliable and that the aircraft had been con-
sistently flying through inclement weather and returning home in
serviceable condition. As for identified shortcomings, the B-2 was
found to need a direct satellite link to national intelligence agencies
to provide its crew with a more current picture of the electronic bat-
tlefield so that the aircraft could be rerouted in near-real time to
avoid any pop-up SAM threats that might have been detected after it
had taken off. It also became apparent, at least to some observers,
that the 509th Bomb Wing’s crew ratio of two two-pilot crews per air-
craft might need to be increased to four crews, or else that provisions
might need to be made for future combat contingencies to allow the
B-2 to operate out of airfields closer to the battlespace in the interest
of reducing mission times.20
Through its consistently effective performance in Allied Force, the

B-2 finally validated the “global reach, global power” concept first ar-
ticulated by the USAF more than a decade earlier. Along with the
B-52 and B-1, it showed the value of combat aircraft that are not de-
pendent on bases near the theater of operations. In addition, its
consistently successful use of JDAM in near-precision attacks against
______________
19Donald B. Rice, “No Stealth to Pentagon’s Bias Against the B-2,” Los Angeles Times,
May 9, 1999.
20David A. Fulghum, “Lessons Learned May Be Flawed,” Aviation Week and Space
Technology, June 14, 1999, p. 205. A serious limiting factor affecting the first of these
suggested solutions is that doubling the B-2’s crew ratio would require either doubling
the number of training sorties and hours flown by the Air Force’s limited B-2 inventory
or reducing the number of sorties and flying hours made available for each B-2 crew
member—to a point where their operational proficiency and expertise would be unac-
ceptably compromised. Alternatively, the Air Force is now taking a close look at using
RAF Fairford, England, and the island bases of Diego Garcia and Guam as forward
staging areas from which to conduct B-2 operations in future regional contingencies
worldwide.
high-priority fixed targets irrespective of weather may, at long last,

have presaged an end to the six-year U.S. habit of routinely resorting
to expensive cruise missiles as a seemingly risk-free way of delivering
precision ordnance. Before the start of Allied Force, the Clinton
administration had expended nearly 800 cruise missiles, all told, in
various punitive attacks against presumed terrorist targets and
against Iraq. At a price penalty of as much as $1.5 million a shot in
sunk costs, that added up to enough to pay for the purchase of 50,000
JDAMs (for a 62:1 cost ratio).21
UAV EMPLOYMENT
Also for the first time in American combat experience, UAVs offered
commanders and planners the frequent advantage of real-time video
imagery without any accompanying danger of aircrew losses. Some
UAVs were flown as low as 1,000 ft above VJ troop positions to gather
real-time imagery, which, in turn, occasionally enabled prompt and
effective attacks by A-10s and F-16s against the often fleeting targets.
Several UAVs were lost when commanders requested closer looks,
forcing the drones to descend into the lethal envelopes of Serb AAA
and man-portable air defense systems (MANPADS). These losses did
not evoke great concern, however, since the UAVs were intentionally
sent out on missions that were known ahead of time to be especially
risky, including highly classified missions to collect and downlink
evidence on Serb atrocities.22
The USAF’s RQ-1A Predator, with a 24-hour endurance capability,

mounted a synthetic-aperture radar that enabled it to track targets
through clouds and thereby augment the two E-8 Joint STARS air-
craft that were operating adjacent to Kosovo out of Germany.
Predator also offered the wherewithal for collecting signals intelli-
gence (SIGINT) through its ability to approach threat emitters more
closely than manned aircraft and to monitor low-power transmis-
______________
21William M. Arkin, “In Praise of Heavy Bombers,” Bulletin of the Atomic Scientists,
July–August 1999. Another 218 U.S. and British TLAMs were fired during Operation
Allied Force.
22“Despite Losses, Backers Say Unmanned Systems Excelling Over Kosovo,” Inside the
Pentagon, June 10, 1999, p. 1.
sions, such as those from cell phones and portable radios operated
by enemy ground troops. 23
The most-advanced Predator was not available when Operation Al-

lied Force began. The USAF initially elected to keep those aircraft at
their home base at Indian Springs near Nellis AFB, Nevada, rather
than commit them to USEUCOM, owing to its reluctance to accept
their delivery from the manufacturer without the accompanying
technical manuals it needed to maintain and effectively operate
them. (Earlier-generation Predators already operating in the theater
were frequently prevented from flying because of their susceptibility
to icing.) 24 The USAF finally sent three advanced Predators to its
UAV facility at the Tuzla airfield in Bosnia. It took more than a week
to get the first Predator airborne over Kosovo, however, because of
undisclosed technical difficulties. In the meantime, USEUCOM and
NATO were obliged to rely on satellites and higher-flying UAVs for
targeting and battle damage assessment (BDA).25
One new procedure demonstrated operationally for the first time in

Kosovo entailed a clever fusion of UAV sensor and specialized com-
mand and control procedures, in which two Predators orbiting at
5,000 ft would provide electro-optical and infrared identification of
mobile targets and a third Predator would then use its laser designa-
tor and mapping software to provide geolocation, after which orbit-
ing A-10s or F-16s could be called in on the detected target. Several
confirmed hits on VJ tanks were made possible by this technique.
Interestingly, Predator was not always used in Operation Allied Force

in the manner in which it was originally designed to be used. In
addition to target search and intelligence collection, the UAV was
also often employed to validate pilot reports of possible SAM or
ground-force targets on the move, since the rules of engagement of-
______________
23John D. Morrocco, David Fulghum, and Robert Wall, “Weather, Weapons Dearth
Slow NATO Strikes,” Aviation Week and Space Technology, April 5, 1999, p. 29.
24“Air Force Reluctant to Deploy All-Weather Predator UAVs to Balkans,” Inside the
Air Force, April 2, 1999, p. 1. Another concern had to do with a larger requirements
debate within the Air Force over whether UAVs developed under a fast-track acquisi-
tion process, as was Predator, should be managed like a more expensive fighter pro-
gram.
25Jane Perlez, “Serbs Try to Empty Disputed Province, NATO Aides Assert,” New York
ten required two sets of eyes on a potential target. As General

Jumper later explained, those who planned and executed the air ef-
fort soon learned that they “had to make forward air controllers out
of what had previously been intelligence collectors.”26 The original
intended Predator mission was to find targets. What happened as
the air war unfolded, however, was that Predator was used instead in
the collateral-damage management loop and sent out to put real-
time eyes on candidate targets that had already been located but not
identified, so as to verify that they were valid military targets. 27
The U.S. Army’s Hunter UAVs operated from the Skopje airfield in
Macedonia, with their first operational mission into Kosovo taking
place on April 4. Hunter imagery was first downlinked to ground
controllers in Skopje and then forwarded either to the CAOC in Vi-
cenza, Italy, or to NATO headquarters in Belgium and to the Pen-
tagon as appropriate.28 Normally used as a corps asset, Hunter in
this instance transmitted real-time video imagery via orbiting satel-
lites and downlinked it directly to the Joint Broadcast System in the
United States, which then transmitted it to the CAOC, making for
only a one-second delay. Its targets were normally objects of tactical
interest against which commanders would not risk a manned air-
craft, such as artillery emplacements and dispersed VJ units in the
KEZ, which had organic self-protection air defense assets. Much like
Predator, Hunter flew whenever the weather allowed. It often would
loiter in the vicinity of hot targets to observe munitions impacts and
provide real-time BDA.29
Both Predator and Hunter operators soon discovered that better sen-
sors were needed for the drones to identify ground targets positively
from above 8,000 ft. They also learned that better integration of UAV
and manned aircraft operations was essential for minimizing the
______________
26“Jumper on Air Power,” p. 42.
27One problem pointed up by this mode of operation was the slow flying speed of the
aircraft. At a maximum airspeed of only 70 nautical miles per hour, Predator typically
required considerable time to get to a previously located target candidate, by which
time the latter may have moved to a new location.
28Elizabeth Becker, “They’re Unmanned, They Fly Low, and They Get the Picture,”
29 Tim Ripley, “Task Force Hunter,” World Air Power Journal, Winter 1999/2000,
p. 122.
danger of midair collisions. As a stopgap toward that end, UAVs were

restricted to operating in specially designated airspace, where they
experienced a heightened likelihood of being shot at because of their
frequency of flight over the same terrain.30 In all, 25 UAVs operated
by all allies went down over the 78-day course of Allied Force as a re-
sult either of enemy action or of mechanical failure. The United
States lost four Predators, eight Hunters (three to infrared SAMs, one
to a radar SAM, and the others for mechanical reasons), and four Pi-
oneers. Germany and France lost a total of six Canadian-built CL-
289 drones and two French Crecerelles, most of them in a single
week.31
After Allied Force ended, General Jumper revealed that had combat
operations continued into the summer, the USAF would have started
employing a new tactic whereby Predators equipped with laser des-
ignators would have been flown under the weather near enemy tar-
gets to designate those targets for LGBs once the latter had been re-
leased by allied fighters flying at safer altitudes above the cloud
cover. Jumper further disclosed that UAVs, having successfully un-
dergone a rigorous operational shakedown over Kosovo, would in the
future be used more in the targeting loop than in the intelligence
collection loop—patrolling aggressively and making the most of their
extended loiter time to seek out and identify hidden targets.32
THE CONTRIBUTIONS OF SPACE

Among the many U.S. and European space systems that were in-
volved in supporting the planning and execution of air attacks, the
most pivotal were classified U.S. satellites that provided imagery
support, including transmissions directly through new National Re-
connaissance Office (NRO) data reception hardware which had been
______________
30David A. Fulghum, “Joint STARS May Profit from Yugoslav Ops,” Aviation Week and
Space Technology, July 26, 1999, p. 74.
31William M. Arkin, “Top Air Force Leaders to Get Briefed on Serbia Air War Report,”
Defense Daily, June 13, 2000, p. 1. For further details on UAV operations, see Lieu-
tenant Commander J. D. Dixon, “UAV Employment in Kosovo: Lessons for the Opera-
tional Commander,” paper submitted to the Naval War College, Newport, Rhode
Island, February 8, 2000.
32David A. Fulghum, “Kosovo Conflict Spurred New Airborne Technology Use,” Avia-
tion Week and Space Technology, August 23, 1999, p. 30.
installed in the 31st Air Expeditionary Wing’s Tactical Integrated

Planning (TIP) center at Aviano Air Base, Italy; Defense Meteorologi-
cal Support Program (DMSP) satellites that provided weather im-
agery down to 1,000-ft resolution; the GPS satellite constellation
which enabled the consistently accurate delivery of JDAMs by B-2s;
and various NRO data relay and SIGINT spacecraft. Other allied
space assets used in Operation Allied Force included the NATO-4
communications satellite, a British Skynet satellite, the French Tele-
sat Syracuse system, U.S. Defense Satellite Communications System
(DSCS) satellites, and ultra-high-frequency (UHF) follow-on satel-
lites.33 After the effort ended, U.S. Space Command estimated that
80 percent of the spaceborne communications used during Op-
eration Allied Force had been transmitted via commercial satellite
systems. 34
At least five notable space success stories came out of the Allied
Force experience. The first was the effective use of the Multisource
Tactical System (MSTS) on the B-52 and B-1, which gave bomber
crews real-time situation awareness updates. The system had existed
before but had never previously been used in combat. The second
major success story was the highly successful use of GPS-guided
munitions described earlier, most notably JDAM on the B-2 and the
Navy’s TLAM II. Third was the use of the Defense Support Program
(DSP) satellite constellation for providing real-time battle damage
indications (BDI) as an input into the BDA process. New procedures
toward that end were created and refined for Allied Force that had
never before been used.35 Fourth, the Hook 112 survival radio was
available for use by U.S. aircrews, making an important new role for
space-enabling systems in CSAR. 36 Finally, command and control
______________
33Craig Covault, “Military Space Dominates Air Strikes,” Aviation Week and Space
Technology, March 29, 1999, pp. 31–32.
34Peter Grier, “The Investment in Space,” Air Force Magazine, February 2000, p. 50.
35On the other hand, cockpit multifunction display videotapes showing successfully
impacting munitions were not used in the BDA process by the Joint Analysis Center at
RAF Molesworth, resulting in numerous revisits to targets that were already known by
attacking pilots to have been struck before to good effect. Conversation with Lieu-
tenant Colonel Ray Dissinger, Aviano AB, Italy, June 12, 2000.
36 The Hook 112 was developed by the Air Force for use between downed aircrew
members and CSAR forces to eliminate a problem presented by the previous survival
radio, which allowed enemy monitors to locate the downed crewmember’s position by
personnel in the CAOC coordinated the tasking of terrestrial intelli-

gence, surveillance, and reconnaissance (ISR) assets—notably the
USAF’s U-2s and RC-135 Rivet Joint electronic intelligence (ELINT)
aircraft—with space-based ISR assets (that is, national satellite sys-
tems) to a level never before achieved in a wartime operational set-
ting.37 In all, reported the USAF’s chief provider of operational space
support to warfighters at all levels, space integration into Allied Force
was “the most extensive seen to date.” But there was still ample room
for further improvement in such areas as space doctrine, better
education regarding the nation’s space capabilities for prospective
users, and better integration of these capabilities into the
contingency plans of air component commanders worldwide.38
_____________________________________________________________
triangulating on the relatively lengthy voice exchanges required to coordinate a rescue
by CSAR teams. The Hook 112 communicates the downed crewmember’s position by
means of an encrypted burst transmission that denies enemy monitors the ability to
triangulate. A GPS receiver incorporated in the Hook 112 automatically transmits the
crewmember’s exact location, along with any coded transmissions the downed
crewmember may wish to communicate. Major General Gary Dylewski, “The USAF
Space Warfare Center: Bringing Space to the Warfighter,” in Peter L. Hays et al., eds,
Spacepower for a New Millennium: Space and U.S. National Security, New York,
McGraw-Hill, 2000, p. 96.
37“Space Support to Operation Allied Force: Preliminary Lessons Learned,” briefing
to the author by Colonel Robert Bivins, director of operations, U.S. Air Force Space
Warfare Center, Schriever AFB, Colorado, February 25, 2000.
38Major General Robert Hinson, commander, 14th Air Force, “Space Doctrine Lessons
from Operation Allied Force,” command briefing, Vandenberg AFB, California,
December 16, 1999.
Chapter Six
FRICTION AND OPERATIONAL PROBLEMS
Although NATO’s use of air power in Allied Force must, in the end, be
adjudged a success, some troubling questions arose well before the
air war’s favorable outcome over a number of unexpected and dis-
concerting problems encountered along the way. Some of those
problems, most notably in the area of what air planners came to call
“flex” targeting of elusive VJ troops on the move in Kosovo, were ar-
guably as much a predictable result of prior strategy choices as a re-
flection of any inherent deficiencies in the air weapon itself.1 Of
more serious concern were identified shortcomings that indicated
needed fixes in the realm of tactics, techniques and procedures, and,
in some cases, equipment. Beyond the problem of locating, identify-
ing, and engaging dispersed and hidden light infantry targets, the
shortcomings arousing the greatest consternation included assessed
deficiencies in SEAD, excessively lengthy information and intelli-
gence cycle time, inadvertent civilian casualties, and some serious
deficiencies in alliance interoperability. Also of special concern were
the many problems spotlighted by the U.S. Army’s plagued deploy-
ment of its AH-64 Apache helicopters to Albania and the full extent of
U.S. global military overcommitment that the Allied Force experi-
ence brought to light.
______________
1The “flex” targeting effort entailed the launching of combat aircraft without specific
assigned target locations and coordinates, although tasked to seek out various classes
of targets, either through free search or upon being directed to a specific area of
known or suspected enemy activity by the CAOC or an airborne forward air controller.
101
FRUSTRATIONS WITH THE SEAD EFFORT

In contrast to the far more satisfying SEAD experience in Desert
Storm, the initial effort to suppress Serb air defenses in Allied Force
did not go nearly as well as expected. The avowed going-in objective
of the SEAD operation was to neutralize as many of Serbia’s SAMs
and AAA sites as possible, particularly its estimated 16 SA-3 LOW
BLOW and 25 SA-6 STRAIGHT FLUSH fire control radars. Another
early goal was to take out or suppress long-range surveillance radars
that could provide timely threat warning to MANPADS operators car-
rying shoulder-fired infrared SAMs like the SA-7.
The Serbs, however, kept their SAMs defensively dispersed and op-
erating in an emission control (EMCON) mode, prompting concern
that they were attempting to draw NATO aircraft down to lower alti-
tudes where they could be more easily engaged. Before the initial
strikes, there were reports of a large-scale dispersal of SA-3 and SA-6
batteries from nearly all of the regular known garrisons. The under-
standable reluctance of enemy SAM operators to emit and thus ren-
der themselves cooperative targets made them much harder to find
and attack, forcing allied aircrews to remain constantly alert to the
radar-guided SAM threat throughout the air war. 2 It further had the
effect of denying some high-risk targets for a time, increasing force
package size, and increasing overall SEAD sortie requirements.
Moreover, unlike in the more permissive Desert Storm operating

environment, airspace availability limitations in the war zone typi-
cally made for high predictability on the part of attacking NATO air-
craft, and collateral damage avoidance considerations frequently
prevented the use of the most tactically advantageous attack head-
ings. The resulting efforts to neutralize the Serb IADS were, accord-
ing to retired U.S. Navy Admiral Leighton Smith, the commander of
NATO forces in Bosnia from 1994 to 1996, “like digging out potatoes
one at a time.”3 The commander of USAFE, General Jumper, later
added that the CAOC could never get NATO political clearance to at-
tack the most troublesome early warning radars in Montenegro,
______________
2Dana Priest, “NATO Unlikely to Alter Strategy,” Washington Post, March 26, 1999.
3Dana Priest, “NATO Pilots Set to Confront Potent Foe,” Washington Post, March 24,
1999.
Friction and Operational Problems 103
which meant that the Serbs knew when attacks were coming most of
the time.4 In other cases, the cumbersome command and control ar-
rangements and the need for prior CAOC approval before fleeting
pop-up IADS targets detected by Rivet Joint or other allied sensors
could be attacked resulted in many lost opportunities and few hard
kills of enemy SAM sites.
Operation Allied Force drew principally on 48 USAF F-16CJs and 30

Navy and Marine Corps EA-6B Prowlers, along with Navy F/A-18s
and German and Italian electronic-combat role (ECR) Tornados, to
conduct the suppression portion of allied counter-SAM operations.
Land-based Marine EA-6Bs were tied directly to attacking strike
packages and typically provided ECM support for missions con-
ducted by U.S. aircraft. Navy Prowlers aboard the USS Theodore Roo-
sevelt supported carrier-launched F-14 and F/A-18 raids and strike
operations by allied fighters. The carrier-based Prowlers each car-
ried two AGM-88 high-speed antiradiation missiles (HARMs). Those
operating out of Aviano, in contrast, almost never carried even a sin-
gle HARM, preferring instead to load an extra fuel tank because of
their longer route to target. This compromise was often compen-
sated for by teaming the EA-6B with HARM-shooting F-16CJs or
Luftwaffe Tornado ECR variants.5
The USAF’s EC-130 Compass Call electronic warfare aircraft was

used to intercept and jam enemy voice communications, thereby al-
lowing the EA-6Bs to concentrate exclusively on jamming enemy
early warning radars. The success of the latter efforts could be vali-
dated by the RC-135 Rivet Joint ELINT aircraft, which orbited at a
safe distance from the combat area. The biggest problem with the
EA-6B was its relatively slow flying speed, which prevented it from
keeping up with ingressing strike aircraft and diminished its jam-
ming effectiveness as a result. On occasion, the jamming of early
warning radars forced Serb SAM operators to activate their fire-
control radars, which in turn rendered them susceptible to being
______________
4General John Jumper, USAF, “Oral Histories Accomplished in Conjunction with Op-
eration Allied Force/Noble Anvil.”
5 Robert Wall, “Sustained Carrier Raids Demonstrate New Strike Tactics,” Aviation
Week and Space Technology, May 10, 1999, p. 37.
attacked by a HARM. Accordingly, enemy activation of SAM fire-

control radars was limited so as to increase their survivability.6
SEAD operations conducted by F-16CJs almost invariably entailed

four-ship formations. The spacing of the formations ensured that
the first two aircraft in the flight were always looking at a threat area
from one side and the other two were monitoring it from the oppo-
site side. That enabled the aircraft’s HARM Targeting System (HTS),
which only provided a 180-degree field of view in the forward sector,
to maintain 100-percent sensor coverage of a target area whenever
allied strike aircraft were attempting to bomb specific aim points
within it. According to one squadron commander, the F-16CJs
would arrive in the target area ahead of the strikers and would build
up the threat picture before the strikers got close, so that the latter
could adjust their ingress routes accordingly. In so doing, the
F-16CJs would provide both the electronic order of battle and the air-
to-air threat picture as necessary. The squadron commander added
that enemy SAM operators got better at exploiting their systems at
about the same rate that the F-16CJ pilots did, resulting in a continu-
ous “cat and mouse game” that made classic SAM kills “hard to come
by.”7
As noted in Chapter Three, only a few SAMs were reported to have

been launched against attacking NATO aircraft the first night. The
second night, fewer than 10 SA-6s were fired, with none scoring a hit.
Later during Allied Force, enemy SAMs were frequently fired in large
numbers, with dozens launched in salvo fashion on some nights but
only a few launched on others. Although these ballistic launches
constituted more a harassment factor than any serious challenge to
NATO operations, numerous cases were reported of allied pilots be-
ing forced to jettison their fuel tanks, dispense chaff, and maneuver
violently to evade enemy SAMs that were confirmed to be guiding.8
______________
6Robert Wall, “Airspace Control Challenges Allies,” Aviation Week and Space Technol-
ogy, April 26, 1999, p. 30.
7Tim Ripley, “Viper Weasels,” World Air Power Journal, Winter 1999/2000, p. 102. The
standard F-16CJ weapons loadout was two AGM-88 HARMs and four AIM-120 ad-
vanced medium-range air-to-air missiles (AMRAAMs).
8Richard J. Newman, “In the Skies over Serbia,” U.S. News and World Report, May 24,
1999, p. 24. It bears noting here that 10 or more pilots operating in a target area might
report an observed SAM shot as ballistic, while the one pilot on whose helmet the
Indeed, the SAM threat to NATO’s aircrews was far more pronounced
and harrowing than media coverage typically depicted, and aggres-
sive jinking and countermaneuvering against airborne SAMs was fre-
quently necessary whenever the Serbs sought to engage NATO air-
craft. The Supreme Allied Commander in Europe, U.S. Army General
Wesley Clark, later reported that there had been numerous instances
of near-misses involving enemy SAM launches against NATO aircraft,
and General Jumper added that a simple look at cockpit display
videotapes would show that “those duels were not trivial.”9 From the
very start of NATO’s air attacks, Serb air defenders also sought to
sucker NATO aircrews down to lower altitudes so they could be
brought within the lethal envelopes of widely proliferated MANPADS
and AAA systems. A common Serb tactic was to fire on the last air-
craft in a departing strike formation, perhaps on the presumption
that those aircraft would be unprotected by other fighters, flown by
less experienced pilots, and low on fuel, with a consequent limited
latitude to countermaneuver.
The persistence of a credible SAM threat throughout the air war

meant that NATO had to dedicate a larger-than-usual number of
strike sorties to the SEAD mission to ensure reasonable freedom to
operate in enemy airspace. In turn, fewer sorties were available for
NATO mission planners to allocate against enemy military and in-
frastructure targets—although the limited number of approved tar-
gets at any one time tended to minimize the practical effects of that
consequence. Moreover, the Block 50 F-16CJ, which lacked the abil-
ity to carry the LANTIRN targeting pod, was never used for night
precision bombing because it could not self-designate targets.
One of the biggest problems to confront attacking NATO aircrews

on defense-suppression missions was target location. Because of
_____________________________________________________________
missile was figuratively guiding would be actively reacting to it. Shortly thereafter, 10
pilots would recover to widely dispersed home bases and report nonthreatening bal-
listic launches, while only one would return with the evidence of a guided shot. This
drove a perception among Allied Force leaders that “most” of the SAM shots observed
were ballistic. Once all the pertinent information was fused and duplicate reporting
was factored out, however, it turned out that a substantial number of SAM launches
(perhaps as many as a third) were guided. Comments on an earlier draft by Hq
USAFE/IN, May 18, 2001.
9Cited in “Ground Troops Lauded,” European Stars and Stripes, August 6, 1999, and
“Jumper on Air Power,” Air Force Magazine, July 2000, p. 41.
Kosovo’s mountainous terrain, the moving target indicator (MTI)

and SAR aboard the E-8 Joint STARS could not detect objects of in-
terest in interspersed valleys that were masked from view at oblique
look angles, although sensors carried by the higher-flying U-2 often
compensated for this shortfall.10 The cover provided to enemy air
defense assets by the interspersed mountains and valleys was a se-
vere complicating factor. Similarly, efforts to attack the internetted
communications links of the Yugoslav IADS were hampered by the
latter’s extensive network of underground command sites, buried
land lines, and mobile communications centers. Using what was
called fused radar input, which allowed the acquisition and tracking
of NATO aircraft from the north and the subsequent feeding of the
resulting surveillance data to air defense radars in the south, this in-
ternetting enabled the southern sector operations center to cue de-
fensive weapons (including shoulder-fired man-portable SAMs and
AAA positions) at other locations in the country where there was no
active radar nearby. That may have accounted, at least in part, for
why the F-16CJ and EA-6B were often ineffective as SAM killers, since
both employed the HARM to home in on enemy radars that normally
operated in close proximity to SAM batteries.11
In all, well over half of the HARM shots taken by allied SEAD aircrews
were preemptive targeting, or so-called PET, shots, with a substantial
number of these occurring in the immediate Belgrade area.12 Many
HARM shots, however, were reactive rather than preplanned, made
in response to transitory radar emissions as they were detected.13
Yugoslavia’s poorly developed road network outside urban areas

may also have worked to the benefit of NATO attackers on more than
a few occasions because enemy SAM operators depended on road
______________
10 Further mitigating this constraint, the limited surveillance range of Joint STARS
caused by interposed ridge lines restricted E-8 operations primarily with regard to
Kosovo, which harbored only a limited SAM threat (only one of the 5 SA-6 regiments
and no SA-2s or SA-3s). Most of the enemy IADS targets were assessed to lie outside
Kosovo. Moreover, the U-2 and Rivet Joint typically performed well and did not suffer
the same problems that sometimes plagued the E-8. Comments on an earlier draft by
Hq USAFE/IN, May 18, 2001.
11Wall, “Airspace Control Challenges Allies.”
13Wall, “Airspace Control Challenges Allies,” p. 30.
transportation for mobility and towed AAA tended to bog down when
driven off prepared surfaces and into open terrain. NATO pilots
therefore studiously avoided flying down roads and crossed them
when necessary at 90-degree angles to minimize their exposure time.
By remaining at least 5 km from the nearest road, they often were
able to negate the AAA threat, albeit at the cost of making it harder to
spot moving military vehicles.
Whenever available intelligence permitted, the preferred offensive

tactic entailed so-called DEAD (destruction of enemy air defense) at-
tacks aimed at achieving hard kills against enemy SAM sites using the
Block 40 F-16CG and F-15E carrying LGBs, cluster bomb units
(CBUs), and the powered AGM-130, rather than merely suppressing
SAM radar activity with the F-16CJ and HARM. 14 For attempted
DEAD attacks, F-16CGs and F-15Es would loiter near tankers orbiting
over the Adriatic to be on call to roll in on any pop-up SAM threats
that might suddenly materialize.15 The unpowered AGM-154 Joint
Standoff Weapon (JSOW), a “near-precision” glide weapon featuring
inertial and GPS guidance and used by Navy F/A-18s, was also effec-
tive on at least a few occasions against enemy acquisition and track-
ing radars using its combined-effects submunitions.16
One problem with such DEAD attempts was that the data cycle time
had to be short enough for the attackers to catch the emitting radars
before they moved on to new locations. One informed report ob-
served that supporting F-16CJs were relatively ineffective in the re-
active SEAD mode because the time required for them to detect an
impending launch and get a timely HARM shot off to protect a striker
______________
14The AGM-130 could be fired from a standoff range of up to 30 nautical miles. It fea-
tured GPS guidance, enhanced by terminal homing via man in the loop through live
video feed data-linked to the attacking aircraft from the guiding weapon.
15The Block 50/52 F-16CJs used for defense suppression were equipped to carry the
AGM-65 Maverick missile, but they did not employ that munition in Allied Force be-
cause the pilots, given their predominant focus on making the most of the AGM-88
HARM, had not sufficiently trained for its use.
16Gelwix, “Oral Histories.” JSOW was employed only infrequently during Allied Force.
Many of the targets assigned to the Navy were inappropriate for attack by the AGM-
154’s cluster-bomb variant because of collateral damage concerns and the lengthy
timelines associated with attacks against mobile targets and with the munition’s lack
of a precise impact timeline. William M. Arkin, “Fleet Praises JSOW, Lists Potential
Improvements,” Defense Daily, April 26, 2000.
invariably exceeded the flyout time of the SAM aimed at the targeted
aircraft. As a result, whenever attacking fighters found themselves
engaged by a SAM, they were pretty much on their own in defeating
it. That suggested to at least some participating aircrews the value of
having a few HARMs uploaded on selected aircraft in every strike
package so that strikers could protect themselves as necessary with-
out having to depend in every case on F-16CJ or EA-6B support.17
The commander of the Marine EA-6B detachment at Aviano com-

mented that there was no single-solution tactic that allied SEAD as-
sets could employ to negate enemy systems. “If we try to jam an
emitter in the south,” he said, “there may be a northern one that can
relay the information through a communications link and land line.
They are fighting on their own turf and know where to hide.”18 The
detachment commander added that Serb SAM operators would peri-
odically emit with their radars for 20 seconds, then shut down the
radars to avoid swallowing a HARM.
In all, more than 800 SAMs were reported to have been fired at NATO
aircraft, both manned and unmanned, over the course of the 78-day
air war, including 477 SA-6s and 124 confirmed man-portable in-
frared missiles (see Figure 6.1 for a depiction of reported enemy SAM
launches by type).19 A majority of the fixed SAMs were fired without
any radar guidance. Yet despite that expenditure of assets, only two
NATO aircraft, an F-117 and an F-16, were shot down by enemy fire,
although another F-117 sustained light damage from a nearby SA-3
detonation and two A-10s were hit by enemy AAA fire but not
downed. 20 There also were two reported cases of short-range
infrared (IR)-guided missiles hitting A-10s, one of which apparently
struck the bottom of the aircraft, defused itself, and bounced off
______________
17Lieutenant Colonel Philip C. Tissue, USMC, “21 Minutes to Belgrade,” Proceedings,
U.S. Naval Institute, September 1999, p. 40.
18Michael R. Gordon, “NATO to Hit Serbs from 2 More Sides,” New York Times, May
11, 1999.
19“AWOS Fact Sheet,” Hq USAFE/SA, December 17, 1999. See also William M. Arkin,
“Top Air Force Leaders to Get Briefed on Serbia Air War Report,” Defense Daily, June
13, 2000, p. 1.
20David A. Fulghum, “Kosovo Report to Boost New JSF Jamming Role,” Aviation Week
and Space Technology, August 30, 1999, p. 22.
RAND MR1365-6.1
Total: 815
Unknown
(26)
IR/MANPAD
(124) SA-3
(188)
SA-6
(477)
Figure 6.1—Enemy SAM Launches Reported
harmlessly. 21 At least 743 HARMs were fired by U.S. and NATO air-
craft against the radars supporting these enemy SAMs (Figure 6.2
provides a detailed breakout of HARM expenditure by target type).22
Yet enough of the Serb IADS remained intact to require NATO
fighters to operate above the 15,000-ft hard deck for most of the air
effort. The main reason for this requirement was the persistent AAA
and MANPADS threat. Although the older SA-7 could be effectively
______________
21“Washington Outlook,” Aviation Week and Space Technology, September 20, 1999,
p. 25.
22“AWOS Fact Sheet,” Hq USAFE/SA, December 17, 1999.
RAND MR1365-6.2
Total: 743
Unknown
(20)
Early warning
radars
(125)
SA-2
(1)
SA-6
(389)
SA-3
(208)
Figure 6.2—HARM Expenditures by Target Type
countered by flares if it was seen in time, the SA-9/13, SA-14, SA-16,

and SA-18 presented a more formidable threat.
In the end, as noted above, only two aircraft (both American) were
brought down by enemy SAM fire, thanks to allied reliance on elec-
tronic jamming, the use of towed decoys, and countertactics to
negate enemy surface-to-air defenses.23 However, NATO never fully
succeeded in neutralizing the Serb IADS, and NATO aircraft operat-
ing over Serbia and Kosovo were always within the engagement en-
velopes of enemy SA-3 and SA-6 missiles—envelopes that extended
______________
23 In all, 1,479 ALE-50 towed decoys were expended by U.S. aircraft during Allied
Force.
to as high as 50,000 ft. Because of that persistent threat, mission

planners were forced to place such high-value ISR platforms as the
U-2 and Joint STARS in less-than-ideal orbits to keep them outside
the lethal reach of enemy SAMs. Even during the operation’s final
week, NATO spokesmen conceded that only three of Serbia’s approx-
imately 25 known mobile SA-6 batteries had been confirmed de-
stroyed.24
In all events, by remaining dispersed and mobile, and activating their

radars only selectively, the Serb IADS operators yielded the short-
term tactical initiative in order to present a longer-term operational
and strategic challenge to allied air operations. The downside of that
inactivity for NATO was that opportunities to employ the classic Wild
Weasel tactic of attacking enemy SAM radars with HARMs while
SAMs were guiding on airborne targets were “few and far be-
tween.”25 The Allied Force air commander, USAF Lieutenant Gen-
eral Michael Short, later indicated that his aircrews were ready for a
wall-to-wall SAM threat like that encountered over Iraq during
Desert Storm, but that “it just never materialized. And then it began
to dawn on us that . . . they were going to try to survive as opposed to
being willing to die to shoot down an airplane.”26 In fact, the survival
tactics employed by Serb IADS operators were first developed and
applied by their Iraqi counterparts in the no-fly zones of Iraq that
have been steadily policed by Operations Northern and Southern
Watch ever since the allied coalition showed its capability against ac-
tive SAM radars during the Gulf war. That should not have come as
any great surprise to NATO planners, and it is reasonable to expect
more of the same as potential future adversaries continue to monitor
U.S. SEAD capabilities and operating procedures and to adapt their
countertactics accordingly.
______________
24Comments on an earlier draft by Hq USAFE/IN, May 18, 2001.
25Tim Ripley, “‘Serbs Running Out of SAMs,’ Says USA,” Jane’s Defense Weekly, June 2,
1999.
26Interview with Lieutenant General Michael Short, USAF, PBS Frontline, “War in Eu-
rope,” February 22, 2000. Serb IADS operators may have been able to trade short-term
effectiveness for longer-term survivability because allied aircraft were typically unable
to find and successfully attack VJ fielded forces and other mobile ground targets. Had
they been able to do so and to kill VJ troops in large numbers, the VJ’s leadership
would have insisted on a more aggressive air defense effort. That would have enabled
NATO to kill more SAMs, but at the probable cost of more friendly aircraft lost. I am
indebted to my RAND colleague John Stillion for this insight.
The dearth of enemy radar-guided SAM activity may also have been
explainable, at least in part, by reports that the Air Force’s Air Com-
bat Command had been conducting information operations by in-
serting viruses and deceptive communications into the enemy’s
computer system and microwave net.27 Although it is unlikely that
U.S. information operators were able to insert malicious code into
enemy SAM radars themselves, General Jumper later confirmed that
Operation Allied Force had seen the first use of offensive computer
warfare as a precision weapon in connection with broader U.S. in-
formation operations against enemy defenses. As he put it, “we did
more information warfare in this conflict than we have ever done
before, and we proved the potential of it.” Jumper added that al-
though information operations remained a highly classified and
compartmented subject about which little could be said, the Kosovo
experience suggested that “instead of sitting and talking about great
big large pods that bash electrons, we should be talking about mi-
crochips that manipulate electrons and get into the heart and soul of
systems like the SA-10 or the SA-12 and tell it that it is a refrigerator
and not a radar.”28 Such pioneering attempts at offensive cyber-
warfare pointed toward the feasibility of taking down SAM and other
defense systems in ways that would not require putting a strike pack-
age or a HARM missile on critical nodes to neutralize them.
During Desert Storm, by means of computer penetration, high-speed

decrypting algorithms, and taps on land lines passing through
friendly countries, the United States was reportedly able to intercept
and monitor Iraqi email and digitized messages but engaged in no
manipulation of enemy computers. During Allied Force, however,
information operators were said to have succeeded in putting false
targets into the enemy’s air defense computers to match what enemy
controllers were predisposed to believe. Such activities also report-
edly occasioned the classic operator-versus-intelligence conundrum
from time to time, in which intelligence collectors sought to preserve
enemy threat systems that were providing them with streams of in-
______________
27David A. Fulghum, “Serb Threat Subsides, but U.S. Still Worries,” Aviation Week and
Space Technology, April 12, 1999, p. 24.
28“Jumper on Air Power,” p. 43.
formation while operators sought to attack them and render them

useless in order to protect allied aircrews.29
Fortunately for NATO, the Serb IADS did not include the latest-
generation SAM equipment currently available on the international
arms market. There were early unsubstantiated reports, repeatedly
denied by the Russian Ministry of Foreign Affairs, that several weeks
before the start of the bombing effort, Russia had provided
Yugoslavia with elements of between six and ten S-300PM (NATO
code-name SA-10) long-range SAM systems, which had been deliv-
ered without their 36D6 Clam Shell low-altitude acquisition radars.30
Had those reports been valid, even the suspected presence of SA-10
and SA-12 SAMs in the enemy IADS inventory would have made life
far more challenging for attacking NATO aircrews. Milosevic
reportedly pressed the Russians hard for such equipment repeatedly,
without success. Deputy Secretary of State Strobe Talbott later
stated that the Yeltsin government had been put on the firmest
notice by the Clinton administration that any provision of such
cutting-edge defensive equipment to Yugoslavia would have had a
“devastating” effect on Russian-American relations.31
All of this raised basic questions about the adequacy of U.S. SEAD
tactics and suggested a need for better real-time intelligence on
mobile enemy IADS assets and a means of getting that information
to pilots quickly enough for them to act on it, as well as for greater
standoff attack capability.32 The downings of both the F-117 and F-
16 were attributed to breakdowns in procedures aimed at detecting
______________
29David A. Fulghum, “Yugoslavia Successfully Attacked by Computers,” Aviation Week
and Space Technology, August 23, 1999, pp. 31–34.
30Zoran Kusovac, “Russian S-300 SAMs ‘In Serbia,’” Jane’s Defense Weekly, August 4,
1999.
31Michael Ignatieff, Virtual War: Kosovo and Beyond, New York, Henry Holt and
Company, Inc., 2000, p. 109.
32For example, the SA-10 and SA-12, now available on the international arms market
for foreign military sale, are lethal out to a slant range of some 80 nautical miles, five
times the killing reach of the earlier-generation SA-3 (David A. Fulghum, “Report Tal-
lies Damage, Lists U.S. Weaknesses,” Aviation Week and Space Technology, February
14, 2000, p. 34). One SA-10/12 site in Belgrade and one in Pristina could have provided
defensive coverage of all of Serbia and Kosovo, as well as threatened Compass Call and
the ABCCC operating outside enemy airspace.
enemy IADS threats in a sufficiently timely manner and ensuring that

pilots did not fly into lethal SAM envelopes unaware of them. Other
factors cited in the two aircraft downings were faulty mission plan-
ning and an improper use of available technology (see below for
more on the F-117 downing). Although far fewer aircraft were lost
during Allied Force than had been expected, these instances pointed
up some systemic problems in need of fixing. As one Air Force gen-
eral observed, “there had to be about 10 things that didn’t go right.
But the central issue is an overall lack of preparedness for electronic
warfare.”33
One of the first signs of this insidious trend cropped up as far back as
August 1990, when half of the Air Force’s ECM pods being readied for
deployment to the Arabian peninsula for Desert Storm were found to
have been in need of calibration or repair. Among numerous later
sins of neglect with respect to electronic warfare (EW) were Air Force
decisions to make operational readiness inspections (ORIs) and
Green Flag EW training exercises less demanding, decisions that nat-
urally resulted in an atrophying of the readiness inspection and re-
porting of EW units, along with a steady erosion of EW experience at
the squadron level. “Now,” said the Air Force general cited above,
“they only practice reprogramming [of radar warning receivers] at
the national level. Intelligence goes to the scientists and says the sig-
nal has changed. Then the scientists figure out the change for the
[ECM] pod and that’s it. Nobody ever burns a new bite down at the
wing.”34
During the years since Desert Storm, the response time for SEAD
challenges has become longer, not shorter, owing to an absence of
adequate planning and to the disappearance of a talent pool of Air
Force leaders skilled in EW. One senior Air Force Gulf War veteran
complained that “we used to have an XOE [operational electronic
______________
33David A. Fulghum, “NATO Unprepared for Electronic Combat,” Aviation Week and
Space Technology, May 10, 1999, p. 35. A thorough and detailed account of the many
problems and concerns identified and highlighted with respect to the USAF’s current
SEAD and electronic warfare repertoire is contained in the summary report of an Air
Force–commissioned study by RAND’s Natalie Crawford and seven senior retired Air
Force general-officer electronic warfare experts, “USAF EW Management Process
Study,” briefing charts, October 1, 1999.
34Fulghum, “NATO Unprepared for Electronic Combat,” p. 35.
warfare] branch in the Air Staff. That doesn’t exist any more. We
used to reprogram [ECM] pods within the wings. They don’t really
do that any more.” During a subsequent colloquium on the air war
and its implications, former Air Force chief of staff General Michael
Dugan attributed these problems to the Air Force’s having dropped
the ball badly in 1990, when it failed to “replace a couple of senior
officers in the acquisition and operations community who [oversaw]
the contribution of electronic combat to warfighting output. The
natural consequence was for this resource to go away.”35
A particular concern prompted by the less-than-reassuring SEAD ex-

perience in Allied Force was the need for better capabilities for ac-
commodating noncooperative enemy air defenses and, more specifi-
cally, countering the enemy tactic whereby Serb SAM operators
resorted to passive electro-optical rather than active radar tracking.
That tactic prompted Major General Dennis Haines, Air Combat
Command’s director of combat weapons systems, to spotlight the
need for capabilities other than relying on radar emissions to detect
SAM batteries, as well as to locate and fix on enemy SAM sites more
rapidly when they emitted only briefly.36 Looking farther down-
stream, one might also suggest that in the long run, the answer is not
to continue getting better at SEAD but rather to move to improved
low-observability capabilities and to the use of UCAVs (unmanned
combat air vehicles), with a view toward rendering SEAD increasingly
unnecessary.
Such concerns have occasioned a growing sense among SEAD spe-

cialists that the management of EW should be taken out of the do-
main of information operations, where it was pigeonholed for con-
venience after the retirement of the EF-111 and F-4G, and returned
to its proper home at the USAF Air Warfare Center at Nellis AFB,
Nevada. As one senior officer complained in this respect, electronic
combat after Desert Storm found itself “buried in with information
operations and information attack. What got lost was the critical is-
sue that EW is a component of combat aircraft survivability.” 37 One
______________
35“Washington Outlook,” Aviation Week and Space Technology, August 23, 1999, p. 27.
36Robert Wall, “SEAD Concerns Raised in Kosovo,” Aviation Week and Space Technol-
ogy, June 26, 1999, p. 75.
37“Washington Outlook,” Aviation Week and Space Technology, June 7, 1999, p. 23.
side result of this neglect of the EW mission by the Air Force was that
maintenance technicians could no longer reprogram quickly (that is,
in 24 hours or less) ECM pods and radar warning receivers to counter
newly detected enemy threats. That problem first arose in 1998,
when several planned U-2 penetrations into hostile airspace had to
be canceled at the last minute because USAF radar warning systems
could not recognize some IADS signals emanating from Iraq and
Bosnia.
Yet another problem highlighted by the IADS challenge presented in

Allied Force was the disconcertingly small number of F-16CJs and
EA-6Bs available to perform the SEAD mission. Aircraft and aircrews
were both stretched extremely thin, even with the modest help pro-
vided by German and Italian Tornado ECR variants. This shortage of
SEAD assets prompted a proposal for backfitting the HARM targeting
system carried by the F-16CJ onto older F-16s and F-15Es. Another
fix suggested for the shortfall in SEAD capability was to begin sup-
plementing existing capabilities and tactics, which rely on the small-
warhead HARM, with PGMs and attack tactics aimed at achieving
hard kills against IADS targets for the duration of a campaign, essen-
tially a very different approach. Most telling of all, the uneven results
of the SEAD experience in Allied Force induced Air Combat Com-
mand to seek an increase in its planned acquisition of new F-16CJs
from 30 to 100.38
THE F-117 SHOOTDOWN

It did not take long for the problems connected with the air war’s
SEAD effort to register their first toll. On the fourth night of air op-
erations, in the first combat loss ever of a stealth aircraft, an F-117
was downed at approximately 8:45 p.m. over hilly terrain near Bu-
danovici, about 28 miles northwest of Belgrade, by an apparent bar-
rage of SA-3s. Fortunately, the pilot ejected safely and, against
formidable odds, was recovered before dawn the next day by a
______________
38“Washington Outlook,” Aviation Week and Space Technology, May 24, 1999, p. 27.
combat search and rescue team using MH-53 Pave Low and MH-60
Pave Hawk helicopters, and directed by a flight of A-10s.39
There was a flurry of speculation afterward as to how such an unex-

pected event might have taken place. Experts at Lockheed Martin
Corporation, the aircraft’s manufacturer, reported that unlike earlier
instances of F-117 combat operations, the missions flown over Yu-
goslavia had required the aircraft to operate in ways that may have
compromised its stealth characteristics. By way of example, they
noted that even a standard banking maneuver can increase the air-
craft’s radar cross-section (RCS) by a factor of 100 or more—and
such turns were unavoidable in the constricted airspace within
which the F-117s were forced to fly.40 Another unconfirmed report
suggested that the RC-135 Rivet Joint aircraft monitoring enemy SAM
activity may have been unable to locate the SA-3 battery that was
thought to have downed the F-117 and may additionally have failed
to relay to the appropriate command and control authorities timely
indications of enemy SAM activity. Lending credence to that inter-
pretation, the commander of Air Combat Command, General
Richard Hawley, commented that “when you have a lot of unlocated
threats, you are at risk even in a stealth airplane.” 41
Although the Air Force has remained understandably silent as to

what confluence of events it believes occasioned the F-117’s down-
ing, press reports claimed that Air Force assessors had concluded, af-
ter conducting a formal postmortem, that a lucky combination of
low-technology tactics, rapid learning, and astute improvisation had
______________
39Although some criticism was voiced afterward as to how CSAR had been shown to
be “broken” because of problems that cropped up during the rescue operation
(apparently, one of the helicopters was forced to disengage, refuel, and penetrate en-
emy airspace a second time before it could find and finally retrieve the downed pilot),
genuine acts of heroism were displayed during the mission. It ended up a brilliant
success and had the welcome effect of turning a propaganda coup for Milosevic al-
most instantly into a propaganda coup for NATO. On the criticism expressed, see
Rowan Scarborough, “Air Force Search and Rescue Operations Called ‘Broken,’”
Washington Times, September 13, 1999.
40James Peltz and Jeff Leeds, “Stealth Fighter’s Crash Reveals a Design’s Limits,” Los
Angeles Times, March 30, 1999.
41“Washington Outlook,” Aviation Week and Space Technology, May 3, 1999, p. 21.
Asked whether the aircraft’s loss was caused by a failure to observe proper lessons
from earlier experience, Hawley added: “That’s an operational issue that is very
warm.”
converged in one fleeting instant to enable an SA-3 not operating in

its normal, radar-guided mode to down the aircraft. Enemy spotters
in Italy doubtless reported the aircraft’s takeoff from Aviano, and
IADS operators in Serbia, as well as perhaps in Bosnia and along the
Montenegran coast, could have assembled from scattered radars
enough glimpses of its position en route to its target to cue a SAM
battery near Belgrade to fire at the appropriate moment. The aircraft
had already dropped one laser-guided bomb near Belgrade, offering
the now-alerted air defenders yet another clue. (The Air Force is said
to have ruled out theories hinging on a stuck weapons bay door, a
descent to below 15,000 ft, or a hit by AAA.)42
At least three procedural errors were alleged to have contributed to

the downing.43 The first was the reported inability of ELINT collec-
tors to track the changing location of the three or four offending SAM
batteries. Three low-frequency Serb radars that at least theoretically
could have detected the F-117’s presence were reportedly not neu-
tralized because U.S. strike aircraft had earlier bombed the wrong
aim points within the radar complexes. Also, F-16CJs carrying
HARMs and operating in adjacent airspace could have deterred the
SA-3 battery from emitting, but those aircraft had been recalled be-
fore the F-117 shootdown.
The second alleged procedural error entailed an EA-6B support

jammer that was said to have been operating not only too far away
from the F-117 (80 to 100 miles) to have been of much protective
value, but also out of proper alignment with the offending threat
radars, resulting in inefficient jamming.
Last was the reported fact that F-117s operating out of Aviano had
previously flown along more or less the same transit routes for four
nights in a row because of a SACEUR ban on overflight of Bosnia to
______________
42Eric Schmitt, “Shrewd Serb Tactics Downed Stealth Jet, U.S. Inquiry Shows,” New
York Times, April 11, 1999. In subsequent testimony before the Senate Armed Services
Committee, Secretary of the Air Force F. Whitten Peters did confirm that the aircraft
had been downed by enemy SAMs. See Vince Crawley, “Air Force Secretary Advocates
C-130, Predators,” Defense Week, July 26, 1999, p. 2.
43See David A. Fulghum and William B. Scott, “Pentagon Gets Lock on F-117 Shoot-
down,” Aviation Week and Space Technology, April 19, 1999, pp. 28–30, and Paul
Beaver, “Mystery Still Shrouds Downing of F-117A Fighter,” Jane’s Defense Weekly,
September 1, 1999.
avoid jeopardizing the Dayton accords. That would have made their
approach pattern into Yugoslav airspace predictable. Knowing from
which direction the F-117s would be coming, Serb air defenders
could have employed low-frequency radars for the best chance of
getting a snap look at the aircraft. Former F-117 pilots and several
industry experts acknowledged that the aircraft is detectable by such
radars when viewed from the side or from directly below. U.S. offi-
cials also suggested that the Serbs may have been able to get brief
nightly radar hits while the aircraft’s weapons bay doors were fleet-
ingly open.
Heated arguments arose in Washington and elsewhere in the imme-

diate aftermath of the shootdown over whether USEUCOM had erred
in not aggressively having sought to destroy the wreckage of the
downed F-117 in order to keep its valuable stealth technology out of
unfriendly hands and eliminate its propaganda value, which the
Serbs bent every effort to exploit.44 Said a former commander of
Tactical Air Command, General John M. Loh: “I’m surprised we
didn’t bomb it, because the standing procedure has always been that
when you lose something of real or perceived value—in this case real
technology, stealth—you destroy it.”45 The case for at least trying to
deny the enemy the wreckage was bolstered by Paul Kaminski, the
Pentagon’s former acquisition chief and the Air Force’s first F-117
program manager during the 1970s. Kaminski noted that although
the F-117 had been operational for 15 years, “there are things in that
airplane, while they may not be leading technologies today in the
United States, are certainly ahead of what some potential adversaries
have.” Kaminski added that the main concern was not that any ex-
ploitation of the F-117’s low-observable technology would enable an
enemy to put the F-117 at greater risk, but rather that it could help
him eventually develop his own stealth technology in due course.46
Reports indicated that military officials had at first considered at-
tempting to destroy the wreckage but opted in the end not to follow
______________
44To bolster their case, some noted that when an F-117 had crashed earlier at an air
show near Baltimore in 1998, the Air Force had thoroughly sanitized the area and
hauled off the wreckage to prevent its most sensitive features from being compro-
mised.
45Vago Muradian, “Stealth Compromised by Not Destroying F-117 Wreckage,” De-
fense Daily, April 2, 1999.
46Ibid.
through with the attempt because they could not have located it
quickly enough to attack it before it was surrounded by civilians and
the media. 47 Those issues aside, whatever the precise explanation
for the downing, it meant not merely the loss of a key U.S. combat
aircraft but the dimming of the F-117’s former aura of invincibility,
which for years had been of incalculable psychological value to the
United States.
PROBLEMS WITH FLEXIBLE TARGETING

Yet another disappointment in the air war’s performance centered
on what turned out to be NATO’s almost completely ineffective ef-
forts to attack mobile VJ forces in the KEZ. By the end of the third
week, despite determined attempts by allied aircrews over the pre-
ceding week, NATO analysts were unable to confirm the destruction
of a single VJ tank or military vehicle, owing to the success of enemy
ground units at dispersing and concealing their armor. That disap-
pointment underscored the limits of conducting air operations
against dispersed and hidden enemy troops in conditions in which
weather, terrain, and tactics all favored the enemy and where no
friendly ground combat presence was on hand to compel those
forces to concentrate and expose themselves. Had Serb comman-
ders any reason to fear a NATO ground invasion, they would have
had little alternative but to position their tanks to cut off roads and
other avenues of attack, thus making their forces more easily tar-
getable by NATO air power. Instead, having dispersed and hidden
their tanks and armored personnel carriers (APCs), Serb army and
paramilitary units were free to go in with just 20 or more troops in a
single vehicle to terrorize a village in connection with their ethnic
cleansing campaign.
______________
47On April 2, the Yugoslav government announced its intention to hand over pieces of
the downed F-117 to Russian authorities. Robert Hewson, “Operation Allied Force:
The First 30 Days,” World Air Power Journal, Fall 1999, p. 18. For the record, it should
be noted that USAF F-15Es were immediately put on alert to destroy the wreckage
with AGM-130s after the F-117 downing was confirmed, but by the time the wreckage
location could be positively determined, CNN was on the scene and collateral damage
issues precluded the attack. Comments on an earlier draft by Hq USAF/XOXS, July 9,
2001.
Indeed, the opportunity to get at fielded enemy ground units with air
power alone had been essentially lost by NATO even before Opera-
tion Allied Force commenced. As General Jumper later recalled,
during the Rambouillet talks in early March 1999, “we watched
40,000 Serbian troops mass north of Kosovo, we watched them infil-
trate down into Kosovo, we watched heavy armor come down into
there, all under the umbrella of the peace conference, and we
weren’t able to react.” 48 Once those forces had completed their
massing on March 15 and had begun a substantial incursion into
Kosovo, any chance for allied air power to be significantly effective
against them promptly disappeared. Once safely dispersed, VJ units
simply turned off the engines of their tanks and other vehicles to save
fuel, hid their vehicles in barns, churches, forests, and populated ar-
eas, hunkered down, and hoped to wait the air effort out. By the end
of April, General Clark frankly conceded that after six weeks of
bombing, there were more VJ, MUP, and Serb paramilitary forces in
Kosovo than there had been when Allied Force began. That attested
powerfully to the latter’s near-total ineffectiveness, at least up to that
point, in halting the Serbian ethnic cleansing rampage throughout
Kosovo.
Once the targeting of enemy troops in Kosovo became a SACEUR

priority at the start of the third week, Yugoslavia was divided into
four large search sectors. The two USAF E-8 Joint STARS aircraft that
had been committed to supporting the air effort were tasked with
searching for ground targets in the KEZ and with providing near-real
time intelligence and targeting information to the CAOC in Vicenza
and to the EC-130 ABCCC. Depending on the possibility of collateral
damage, Joint STARS was sometimes cleared to communicate di-
rectly to airborne FACs and to direct NATO strikes against fleeting
targets of opportunity, with the goal of getting target information and
coordinates to orbiting strike aircraft within minutes.49
______________
49Robert Wall, “Joint STARS Changes Operational Scheme,” Aviation Week and Space
Technology, May 3, 1999, pp. 25–27.
Before long, three broad approaches to what came to be called “flex”

targeting emerged for prompt employment against mobile VJ and
MUP forces operating in Kosovo and against pop-up IADS assets
deployed in Serbia. In the first, called “alert flex” targeting, combat
aircraft were apportioned from the very outset as designated “flex”
sorties in the ATO and reserved for launch on short notice against
any pop-up targets that might be detected and identified within the
ATO cycle. Initially, such designated aircraft were kept on ground
alert. Later in the operation, they were placed on airborne tanker
alert, which reduced their response times by as much as two hours.
The second approach entailed redirecting aircraft already en route to

preplanned fixed targets. Strikers would be diverted either to alter-
nate high-value fixed targets in Serbia or to recently detected mobile
targets in Serbia or Kosovo. Because of the large number of NATO
fighters already preapportioned and available on call for use as alert
flex assets, however, such en route diversions occurred only rarely.
All three heavy bombers (the B-52s, B-1s, and B-2s) were also
diverted to new targets on occasion, requiring real-time changes in
their preplanned ingress routes.
The third category of flexible targeting involved dedicated sorties

launched into holding orbits for on-call attacks against detected
mobile VJ forces in Kosovo after the KEZ was declared on Day 20 of
the air effort. This approach, which evolved progressively over time,
entailed the use of F-16s, A-10s, or Tornados serving as airborne for-
ward air controllers. Their FAC-qualified pilots would search for
ground targets in predesignated kill boxes, attempt a visual identifi-
cation of any suspected target candidates, and assess the potential
for collateral damage after determining that the target candidates
were valid. Depending on the prevailing rules of engagement, the
FAC pilots would first request ABCCC or CAOC approval to attack the
target and then, upon being cleared to release weapons, would drop
their munitions on the approved target while directing their wing-
men to drop on adjacent targets. In the event that multiple targets
were detected and approved, additional strike aircraft would be
called in if they were close at hand. Because NATO had no fielded
ground forces in the combat zone, the FACs could not request
ground assistance and were on their own in locating and identifying
mobile targets.
As noted earlier, a major problem that inhibited the effectiveness of

Joint STARS in support of these missions was Kosovo’s mountainous
terrain, which required the aircraft to fly unusually close to enemy
territory so its sensor operators could look into valleys and minimize
the enemy’s opportunities to take advantage of terrain masking.
Even then, the high ridgelines often made it impossible for Joint
STARS crews, from their standoff orbits, to peer into some valleys
where VJ forces were thought to have been concentrated. Joint
STARS also had only a limited ability to detect and monitor ground
targets in dense woods and built-up areas. Because of these con-
straints, NATO had little by way of wide-area airborne surveillance
and cueing of the sort that had made coalition operations against en-
emy ground forces so effective in Desert Storm. That deficiency
placed a doubly high premium on hitting enemy ground-force tar-
gets as they moved into open areas and were visually detected by
airborne FACs. It also, in effect, ceded the tactical initiative to VJ
forces, since the latter could decide when and where to reposition
themselves. The net result was a need for large numbers of combat
aircraft continuously orbiting over the KEZ but producing little tacti-
cal return, compounded—indeed, largely caused—by the absence of
a NATO ground threat to force enemy troops into more predictable
patterns of behavior.50
The performance of Joint STARS against dispersed and hidden en-

emy forces was less than satisfactory not only because of the con-
straints described above, but also because of an unfortunate failure
by air operations managers to make the most of the aircraft’s inher-
ent capabilities for supporting counterland operations. That failure
partly reflected a continuing slowness on the part of the U.S. Air
Force to develop and institutionalize a detailed appreciation for how
land forces operate and, in turn, to acquire the conceptual where-
withal that is essential for making air power more effective in defeat-
ing those forces. Surprisingly little progress was registered by the Air
Force over the nine years since Desert Storm in developing a concept
of operations for using Joint STARS in a surveillance and control
team that also includes AWACS, Rivet Joint, airborne FACs, and
UAVs, all working as a synergistic collective against elusive enemy
ground forces.
______________
50I am indebted to my colleague John Stillion for developing these points.
As one telling testament to this failure, the inclusion of Joint STARS

in the air war’s equipment roster had been requested by the Army,
not by the Air Force.51 Because of the predominant USAF focus on
attacking fixed infrastructure targets, few in the Air Force fully ap-
preciated the E-8’s capability for providing wide-area, all-weather
standoff coverage of the KEZ and its resultant ability to provide
USEUCOM’s and NATO’s operational-level commanders with real-
time situation awareness regarding the status and activity of VJ
forces. It took days for Joint STARS even to be included in the ATO.
Once there, the aircraft was typically thought of as a surveillance
platform operating in the service of the intelligence community,
rather than as a strike support asset working to provide direct and
immediate assistance to NATO aircrews conducting flexible targeting
missions. With the right teaming, connectivity, and practice, the use
of Joint STARS to cue UAVs might have reduced, if not eliminated,
the “searching-through-a-soda straw” problem, lessened UAV expo-
sure to hostile fire, and helped maintain tactical surprise for NATO
aircrews engaged in the search for VJ targets of opportunity. No
measures of that sort, however, were attempted until quite late in Al-
lied Force.
Yet another complicating influence on the air effort’s attempts

against dispersed and hidden enemy forces stemmed from the
command and control arrangements that had been hastily cobbled
together at the operational and tactical levels once it became clear
that NATO was committed to an air war for the long haul. Although
the CAOC eventually worked out a means of using real-time imagery
to detect fielded VJ forces in the KEZ and to “flex” allied air assets to
attack those newly developed targets in an orderly fashion, those
doing the “flex” decisionmaking during the first half of Allied Force
did so with no apportionment or targeting guidance whatever. As
one expert observer noted, “if the detected target was militarily sig-
nificant, it was struck, regardless of [General Short’s] priorities or in-
tentions. There was no link to an assessment mechanism, so that
once a target was struck, there was no way to link it to what unit it
had been associated with, so no effective degradation was re-
______________
51Personal communication to the author from Price Bingham, Northrop Grumman
Corporation, Melbourne, Florida, December 20, 1999.
corded.”52 As a result, combat aircraft were sometimes diverted

from scheduled ATO targets of clear operational significance to at-
tack “flex” targets of highly dubious tactical, let alone operational or
strategic, worth. Moreover, owing to the absence of any feedback
mechanism, aircraft were often committed against targets that had
already been successfully struck, forcing the CAOC either to re-role
aircraft on short notice or else to expose aircrews needlessly to en-
emy IADS threats a second time. For most of the air war, roughly half
of General Short’s available surface-attack sorties were committed
against targets in the KEZ. Of those, a significant percentage were
“flexed” in this haphazard manner.53
Weather was still another complicating factor in the effort against

dispersed VJ forces. From the 15,000-ft altitude floor above which
NATO aircrews typically operated, the cloud cover over Kosovo was
greater than 50 percent for more than 78 percent of the air war’s du-
ration. That allowed unimpeded strike operations on only 24 of the
air war’s 78 days. The impact of these conditions on the flexible tar-
geting effort was considerable. In all, 3,766 planned sorties, includ-
ing 1,029 designated close air support sorties, had to be canceled be-
cause of weather.
Even on clear days, another factor preventing the kill box system
from being as effective as it might otherwise have been was the tight
rules-of-engagement regime that had been imposed after the
Djakovica incident (see below), in which more than 60 ethnic Alba-
nian refugees were reportedly killed in an attack by USAF F-16s
against what was thought to have been a VJ troop convoy. These re-
strictions had a far greater inhibiting influence on the effectiveness of
NATO’s flexible targeting efforts than the oft-cited 15,000-ft altitude
floor which NATO’s aircrews had been directed to observe. Unless
an object of interest was clearly determined to be a valid military tar-
get, such as a VJ tank operating in the open, pilots had to get clear-
ance for any attack from the CAOC, with General Short himself often
______________
52Lieutenant Colonel L. T. Wight, USAF, “What a Tangled Web We Wove: An After-
Action Assessment of Operation Allied Force’s Command and Control Structure and
Processes,” unpublished paper, no date, p. 12. Colonel Wight was a member of the
C-5 Strategy Cell at the CAOC.
53Ibid.
making the decision after checking second sources like real-time

UAV video feed. Because of the delays created by these and similar
hurdles, orbiting NATO aircraft often ran low on fuel before being
cleared to drop their weapons and accordingly were forced to leave
the area in search of a tanker.54
Last, and perhaps as decisive as any single other factor, VJ forces ag-
gressively avoided making themselves easy targets for NATO air at-
tacks. Indeed, digging in and hunkering down for defensive attrition
warfare had lain at the heart of Yugoslav operational doctrine ever
since the days of partisan operations against the Wehrmacht in
World War II. Whenever General Clark would say, “You’ve got to get
them in their assembly areas,” the reply typically was: “These guys
aren’t assembling!”55 RAF Harrier GR. Mk 7 pilots operating in kill
boxes over Kosovo reported that “there was nothing moving around
at all during the daytime,” adding that when Clark “got up and said
knocking out five tanks was a good day for NATO, he [was] telling it
straight. On some days we couldn’t find any tanks.”56 Even with the
aid of binoculars, the ground below often seemed devoid of life to
NATO aircrews orbiting overhead at 15,000 ft. This was the pre-
dictable result of trying to engage an enemy who had no need to
shoot, move, or expose his position, thanks to the absence of a cred-
ible NATO ground threat.
To be sure, there were some notable bright spots in NATO’s air effort
against VJ forces in Kosovo. To cite one example, in those rare in-
stances in which enemy armor and other targets exposed themselves
to attack from the air, the upgraded AGM-65G2 Maverick air-to-
ground missile generally performed very effectively. The effective-
ness rate for older Mavericks was lower, but still reportedly higher
than 90 percent. 57 Also, both U-2 imagery and pictures provided by
the Navy’s F-14 equipped with TARPS (Tactical Air Reconnaissance
______________
54Tim Ripley, “Harriers over the Kosovo ‘Kill Boxes,’” World Air Power Journal, Winter
1999/2000, p. 100.
56Ripley, “Harriers over the Kosovo ‘Kill Boxes.’”
57Robert Wall, “Maverick Fix Tested in Kosovo,” Aviation Week and Space Technology,
September 6, 1999, pp. 88–89.
Pod System) later proved useful to the CAOC in what the Cohen-
Shelton after-action report to Congress called “several” instances in-
volving the rapid retargeting of NATO aircraft to new targets.58
To cite another notable example, the two Marine F/A-18D squadrons

that deployed to the former Warsaw Pact airfield at Taszar, Hungary,
late in the air war played an active part in the effort against enemy
forces in the KEZ.59 For the first time on a large scale in combat, the
F/A-18D aircrews, along with NATO pilots flying other combat air-
craft types, made heavy use of night-vision goggles with compatible
internal and external lighting modifications, thus enabling multi-
aircraft formations and simultaneous night bomb deliveries.60 Some
F/A-18Ds also carried the internally mounted Advanced Tactical
Aerial Reconnaissance System (ATARS). Still in operational eval-
uation as Allied Force began, the system provided digital, multispec-
tral target images with its SAR and medium-altitude electro-optical
(EO) imagery as a backup to pictures from other ISR sources, with a
real-time connection to ground receiver stations. It figured promi-
nently in both targeting and BDA activities.61
In a typical night F/A-18D flexible targeting mission (which might

last as long as six hours, with four inflight refuelings), the C-130
ABCCC would pass to orbiting Marine fighters the grid coordinates of
a VJ artillery position detected by the TPQ-36 and TPQ-37 counter-
battery radars attached to the U.S. Army’s Task Force Hawk in Alba-
nia. An airborne FAC in an OA-10 would then illuminate the target
______________
58 Secretary of Defense William S. Cohen and Chairman of the Joint Chiefs of Staff
General Henry H. Shelton, Kosovo/Operation Allied Force After-Action Report, Wash-
ington, D.C., Department of Defense, Report to Congress, January 31, 2000, p. 58.
59The airfield itself offered an 8,200-ft runway and a tactical air navigation (TACAN)
system enabling the aircraft to fly instrument approaches, but it lacked a ready com-
munications link to the CAOC in Vicenza and also needed more fuel trucks, as well as
runway arresting gear in the event of wet runways and aircraft emergencies. The latter
were shipped in and quickly became a welcome presence because high-gross-weight
landings in heavy rain proved to be routine.
60As one downside aspect worth noting in this respect, numerous aircrews later indi-
cated that night-vision goggles often provided them with too much information be-
cause they were capable of picking up infrared events as far as 100 miles away.
61 For further details, see Margaret Bone, “Kodak Moments in Kosovo,” The Hook,
Spring 2000, pp. 29–31.
location with flares and call in a two-plane section of F/A-18Ds to be

available on short notice to attack it. In so doing, the OA-10 FAC, in
effect, performed reconnaissance by fire. When shot at in return, the
FAC would determine the source of fire to be hostile, and the F/A-
18Ds would then be cleared to drop 500-lb Mk 82 bombs on it, which
would generally stop the artillery fire for the rest of the night.62 It
was said that the greatest frustration for all NATO aircrews flying
combat missions was to be orbiting over the KEZ night after night,
for as long as six hours interspersed with multiple inflight refuelings,
only to be called in at long last by an airborne FAC and cleared to at-
tack a reported VJ tank that was no longer there.63
Owing in large part to such operations, at least those that produced

recognizable combat results, NATO’s effort to engage dispersed and
hidden enemy forces in the KEZ was not a complete waste of time
and assets. For one thing, VJ commanders knew all too well that as
the weather began steadily improving with the onset of summer, any
effort on their part to conduct large-scale operations against either
the KLA or civilian ethnic Albanians would put them at extremely
high risk of being attacked. Moreover, General Short reported in late
May that the newly focused attacks against the VJ’s 3rd Army in
Kosovo were beginning to register discernible effects. He went on to
predict that “if we do this for two more months, we will either kill this
army in Kosovo or send it on the run.”64
Taken as a whole, however, NATO’s effort to attack enemy ground

units in the KEZ was essentially a failure, the full extent of which be-
came apparent only after the air war was over. To the very end, Short
doubted that focusing exclusively, or even primarily, on elusive VJ
forces in Kosovo would be enough to swing the desired outcome. He
also placed little stock in claims emanating from NATO headquarters
that the VJ was being progressively weakened by the air attacks. On
that latter point, he observed that the only things that mattered were
______________
62Tissue, “21 Minutes to Belgrade,” pp. 38–40.
63Conversation with Major General P. J. M. Godderij, deputy commander in chief,
Royal Netherlands Air Force, Scheveningen, the Netherlands, June 7, 2000.
64William Drozdiak, “Air War Commander Says Kosovo Victory Near,” Washington
Post, May 24, 1999.
that army’s ability to move and its willingness to fight, and that both
of those remained decidedly intact.65
In the first detailed official rundown of the air war’s accomplish-

ments as Allied Force approached its midpoint, the limited effects of
NATO’s bombing attempts against enemy forces in Kosovo were un-
derscored by the frank admission that the VJ still retained 80 to 90
percent of its tanks. 66 Later, on May 19, NATO spokesman Major
General Walter Jertz claimed more optimistically that one-third of all
VJ tanks and artillery in Kosovo had been destroyed.6 7 As the
bombing effort drew to a close, NATO was similarly claiming that it
had taken out more than one-quarter of the VJ’s tanks and APCs de-
ployed in Kosovo. Britain’s chief of the defense staff, General Sir
Charles Guthrie, further reported that more than 30 percent of the
VJ’s artillery and mortar pieces had been destroyed by NATO attack-
ers.68
In its final tally as Operation Allied Force ended, the U.S. Defense
Department settled on 700 out of 1,500 tanks, APCs, and artillery
pieces destroyed altogether in Kosovo.69 More specifically, General
Shelton announced in an early postwar briefing that NATO attacks
had destroyed “around 120 tanks, about 220 armored personnel car-
riers, and up to 450 artillery and mortar pieces.” However, nothing
like a matching number of hulks was found by allied inspectors after
Allied Force ended. During their withdrawal, VJ troops took hun-
dreds of tanks, artillery pieces, and APCs out of Kosovo. They also
seemed spirited and defiant rather than beaten.70 The VJ’s com-
______________
65William M. Arkin, “Limited Warfare in Kosovo Not Working,” Seattle Times, May 22,
1999.
66Paul Richter, “Milosevic War Machine Has a Lot of Fight Left,” Los Angeles Times,
April 29, 1999.
67Robert Hewson, “Allied Force, Part II: Overwhelming Air Power,” World Air Power
Journal, Winter 1999/2000, p. 113.
68 Michael Evans, “Serb Army Talks of Peace as Armor Takes a Pounding,” London
69Rowan Scarborough, “Pentagon Intends to Issue Final Count of Serbian Losses,”
Washington Times, July 9, 1999.
70Over the course of the 11-day Serb withdrawal, NATO observers counted 220 tanks,
300 APCs, and 308 artillery pieces being loaded onto trucks and transporters, along
with hundreds of other vehicles and assorted military equipment. Steven Lee Myers,
“Damage to Serb Military Less Than Expected,” New York Times, June 28, 1999.
mander in chief, General Dragoljub Ojdanic, claimed after the war

that only 524 Yugoslav soldiers had been killed, in marked contrast to
NATO’s estimate of thousands.71
After the dust settled in early June, a preliminary NATO postmortem

concluded that the air war had had almost no effect on VJ operations
in Kosovo. In an after-action briefing to senior Pentagon officials,
the commander of Joint Task Force Noble Anvil, Admiral James Ellis,
confirmed that NATO air operations were effective against VJ armor
only after the KLA launched its offensive, forcing defending VJ troops
to uncover and mass their armor and mechanized forces.72 NATO
initially claimed after the air war ended that it had disabled 150 of the
estimated 400 VJ tanks in Kosovo. General Clark later scaled back
that number to 110, after having determined that many tanks as-
sumed to have been destroyed had, in fact, been decoys that the VJ
had skillfully fielded in large numbers.73
Not only did the Serbs make successful use of tank decoys made out
of tetra-pak milk carton material, they also positioned wood-burning
stoves with their chimneys angled to make them look like artillery
pieces. In some cases, water receptacles were found in the decoys,
cleverly placed there to heat up under the sun to help replicate the
infrared signature of a vehicle or hot artillery tube.74 One source
spoke of cockpit display videotapes showing targets with every ap-
pearance of being tanks collapsing instantly upon being hit. In addi-
tion, the Serbs made heavy and frequently effective use of smoke
generators to protect targets against LGBs. After the air war ended,
______________
71“Yugoslav Army Lost 524 Soldiers, Top General Says,” International Herald Tribune,
July 22, 1999.
72Briefing by Admiral James O. Ellis, USN, commander in chief, U.S. Naval Forces,
Europe, and commander, Allied Forces Southern Europe and Joint Task Force Noble
Anvil, no date given.
73Joseph Fitchett, “NATO Misjudged Bombing Damage,” International Herald Tri-
bune, June 23, 1999. General Jumper dismissed criticisms from some that expensive
U.S. precision munitions had been wasted on decoys. Declaring that U.S. forces had
had “plenty of bombs for decoys,” he noted that what appeared to be legitimate tar-
gets were immediately attacked so that aircrews would not loiter over target areas try-
ing to distinguish real targets from decoys and exposing themselves needlessly to en-
emy fire. Aviation Week and Space Technology, September 20, 1999, p. 25.
74Paul Richter, “U.S. Study of War on Yugoslavia Aimed at Boosting Performance,” Los
Angeles Times, July 10, 1999.
site-survey teams that went in on the ground in Kosovo and inter-

viewed witnesses discovered that VJ forces had buried many of their
missile launchers, covered fuel trucks with rugs, and disguised tanks
as haystacks and armored vehicles as trees.
The subsequent, and putatively definitive, after-action report on Al-

lied Force submitted to Congress by Secretary Cohen and General
Shelton in the summer of 1999 claimed valid strikes on 93 enemy
tanks, 153 APCs, 339 other military vehicles, and 389 artillery and
mortar pieces.75 Those downwardly revised estimates came on the
heels of the findings by a munitions effectiveness assessment (MEA)
team of 67 operators and intelligence experts, made up mostly of
USAF officers, who went into Kosovo at Clark’s behest to comb the
country, both by helicopter and on foot, in an on-site survey of all
actual DMPIs attacked. The team’s specific mission was to perform
an assessment of attacks undertaken against mobile targets in the
Presevo Valley region of Kosovo by cross-referencing on-scene ob-
servations and conversations with witnesses on the ground against
available cockpit display videotapes, imagery intelligence, signals in-
telligence, human intelligence, and interviews with airborne FACs
who had been operating near the target area at the time of the at-
tacks.76
The team’s initial conclusion from that assessment was that “only a
handful” of enemy tanks, APCs, and artillery pieces could be deter-
mined to have been catastrophically damaged by air attacks. 77 Al-
though the team succeeded in investigating some 60 percent of
NATO’s claimed hits on mobile targets in the KEZ, it confirmed only
14 tanks, 18 APCs, and 20 artillery pieces as destroyed for sure. A
later assessment conducted by USAFE’s office of studies and analy-
______________
75Cohen and Shelton, After-Action Report, pp. 84–85.
76As the team’s concept of operations clearly stipulated, the mission objective was to
“determine Allied Force munition effectiveness by selective examination of fixed and
mobile target sets within Kosovo [and to] evaluate and record physical and functional
target damage and precise weapons impact locations and characteristics, with em-
phasis on precision and near-precision air-dropped munitions.” The concept of oper-
ations further stipulated that validation of NATO’s air campaign, target set, BDA, and
rationale for specific target selection were “beyond the scope of this survey.” Docu-
mentation provided to the author by Hq USAFE/SA, May 2, 2001.
77 Tim Butcher and Patrick Bishop, “NATO Admits Air Campaign Failed,” London
Daily Telegraph, July 22, 1999.
sis, using the team’s findings as one important input, reported 93

tanks and 153 APCs as having been struck altogether, the same num-
bers noted above that were cited later by Secretary Cohen and Gen-
eral Shelton. Many of those claimed hits, however, were validated by
only a single source of evidence, such as a cockpit display videotape
or an infrared event detected by DSP satellites.78 In the later after-
math of Allied Force, on-site surveys of bomb damage effects by
KFOR observers and other inspectors further confirmed that NATO’s
attacks against VJ forces had accomplished far less than had initially
been assumed, notably including at Mount Pastrik.79
These seeming discrepancies led some air war critics to charge that
NATO and the U.S. Defense Department were engaging in a blatant
cover-up of allied air power’s poor performance against VJ forces in
Kosovo to avoid being embarrassed by the paltry numbers the in-
spection team had produced. That criticism turned out, however, to
have been overblown for two reasons. First, the cover-up charge was
misdirected, in that it was based entirely on a leaked draft report by
USAFE’s inspection team that went to Kosovo earlier in the summer
of 1999. That draft report, dated August 3, 1999, and titled “Op-
eration Allied Force: Munitions Effectiveness Assessment, Vol. II:
Mobile Targets,” documented information collected in Kosovo and
elsewhere by the MEA working group tasked with looking into
mobile enemy targets. That effort was undertaken not to account for
successful strikes, but rather to determine what equipment remained
at the attacked sites. The freshest of the attacked sites visited was
four weeks old, and some were only visited for the first time three
months after the attacks.
All told, the USAFE team came across 14 tank carcasses and the hulks
of 12 self-propelled artillery vehicles, which could have looked like
tanks from the air and been reported as such in post-strike pilot
mission reports. That added up to 26 confirmable “tanks” suffering
sufficiently catastrophic damage from NATO air attacks to be written
off and abandoned by departing VJ forces. Cross-referencing pilot
reports with corroborating evidence from other sources, the USAFE
______________
78John Barry, “The Kosovo Cover-Up,” Newsweek, May 15, 2000, p. 23.
79Richard J. Newman, “The Bombs That Failed in Kosovo,” U.S. News and World Re-
port, September 20, 1999.
studies and analysis staff later documented presumed successful

strikes on 93 tanks, 153 APCs, and 389 artillery pieces. It further doc-
umented another 60 instances of attacks on tanks that were believed
to have been successful but that could not be validated because of
the stringent criteria it had been given by SACEUR. As explained in
SACEUR’s subsequent strike assessment briefing at NATO headquar-
ters, 26 tanks could be categorized as “confirmed catastrophic kills,”
based on physical information actually gathered on the ground in
Kosovo. The remainder of the 93 reported tank kills were categorized
as “assessed strikes,” which meant, in effect, that there were indica-
tions suggesting that a weapon may have hit a valid target.80
Air warfare professionals, notably including the USAF chief of staff,

General Michael Ryan, have readily acknowledged since the end of
Allied Force that the problems encountered by the operation’s flexi-
ble targeting effort outlined above reflected real challenges for the
effective application of air power posed by such impediments as
trees, mountains, poor weather, and an enemy ground force permit-
______________
80Stephen P. Aubin, “Newsweek and the 14 Tanks,” Air Force Magazine, July 2000, pp.
59–61. As USAFE’s director of studies and analysis, Brigadier General John Corley,
who directed that assessment, explained afterward during a Pentagon press briefing,
“if a pilot claimed that he had attacked a tank at a given [location], we would go to that
location and . . . begin to survey that exact site. If what we had was . . . multiple
sources to confirm what had been claimed, then we would put that into a successful
strike category. Let me give you an example. If we went to one of those desired mean
points of impact and we found a bomb crater and we found shrapnel and oil down in
the bottom of that bomb crater, then we would take a digitized photo of that crater
and we would note that there would be earth scarring, as if some very heavy piece of
equipment had been dragged from that bomb crater out to a road. Then we would
compare that with both before and after imagery. You might have, for example, a
[satellite] image showing a tank in a tree line. You may go and take a look at the
cockpit video which shows that tank at that exact set of coordinates with a munition
impacting it. . . . You may then go back and discover a piece of U-2 film afterward
showing a damaged tank. You may then find out that an airborne forward air
controller who had flown specifically over this area day in and day out would report
that approximately two to three days later, whatever had been there was now gone
from that location. We further wound up with some information whereby we saw
bomb-damaged and destroyed equipment loaded on board flatbed trucks being taken
out of Kosovo, headed back north into Serbia. So as you begin to look at all those
sources of information, those multiple layers worth . . . in concert, and if we had
multiple pieces of evidentiary information, we would confirm a successful strike. And
that was the difference between the 26 and the 93. If we could not confirm with
multiple sources, we did not claim a successful strike.” News briefing, Office of the
Assistant Secretary of Defense (Public Affairs), the Pentagon, Washington, D.C., May 8,
2000.
ted the luxury of dispersing and hiding rather than concentrating to

maneuver to accomplish its mission.81 The Cohen-Shelton report to
Congress frankly admitted that the problems encountered with flex-
ible targeting of VJ forces in Kosovo pointed up continued shortfalls
in the nation’s ability to meet “the difficult challenge of rapidly tar-
geting enemy forces and systems that can move and hide fre-
quently.”82 On that discomfiting point, U.S. and NATO defense offi-
cials had nothing whatever to hide and covered nothing up.
Second, and perhaps more important, although it was clearly essen-

tial for NATO to maintain constant pressure on VJ and MUP forces
deployed in Kosovo and to bend every reasonable effort to suppress
their freedom to operate at will against the ethnic Albanians, the
majority of the combat sorties that SACEUR insisted be devoted to
finding and attacking enemy forces in the KEZ arguably entailed a
waste of munitions and other valuable assets. That perspective was
pithily expressed by the vice chairman of the Joint Chiefs of Staff,
USAF General Joseph Ralston, who later went on to replace Clark as
SACEUR: “The tank, which was an irrelevant item in the context of
ethnic cleansing, became the symbol for Serb ground forces. How
many tanks did you kill today? All of a sudden, this became the mea-
sure of merit that had nothing to do with reality.” 83 When General
Jumper, on being pressed later by reporters for an honest account of
how many tanks NATO had actually destroyed, replied simply
“enough,” he was telling the truth. The marginality of the tank issue
to what really mattered in Allied Force was perhaps most convinc-
ingly explained by Brigadier General Daniel Leaf, commander of the
31st Air Expeditionary Wing at Aviano, when he declared in the im-
mediate wake of the cease-fire that “counting tanks is irrelevant. The
fact is they withdrew, and while they took tanks with them, they re-
turned to a country whose military infrastructure has been ruined.
______________
81Indeed, in its interim report on the Kosovo air effort, the USAF expressly conceded
that “shortfalls remain . . . in the USAF’s ability to locate and attack moving armor and
other ground forces in poor weather. The Air Force needs to continue to develop and
improve its ability to do this.” The Air War Over Serbia: Aerospace Power in Operation
Allied Force, Washington, D.C., Hq United States Air Force, April 1, 2000, p. 53.
82Cohen and Shelton, After-Action Report, p. 56.
83Dana Priest, “Tension Grew with Divide in Strategy,” Washington Post, September
21, 1999.
They’re not going to be doing anything with those forces for a long
time.”84
True enough, a demonstrable record of effective performance by the

attacks against VJ tanks may well have been regarded at the time as
being of crucial importance toward vindicating SACEUR’s stress on
attacks against dispersed and hidden enemy forces in Kosovo. Yet
viewed in hindsight, the number of tanks taken out in the air war
was, and remains, an issue of only scant pertinence to the opera-
tion’s ultimate outcome. Not only that, getting into the tank-count-
ing business in the first place made for a largely self-inflicted wound
by the Department of Defense, SACEUR, and NATO. In the end, all
the to-ing and fro-ing over how many enemy tanks were taken out by
NATO was mainly of academic interest, since air operations in the
KEZ were, by all indications, not a determining factor affecting Milo-
sevic’s ultimate decision to capitulate.85 The KLA had been elimi-
nated entirely as a tactical consideration by superior VJ strength.
Moreover, notwithstanding more than two months of continual
NATO bombing, the VJ lost few personnel to hostile fire, retained its
command and control and resupply apparatus throughout the air
effort, and continued to conduct ethnic cleansing forays until the last
day of the air war, even though it did put itself at risk whenever its
units exposed themselves to attack from the air. At bottom, NATO’s
failure to perform better than it did against enemy ground units in
the KEZ was as much a result of the strategy chosen by its leaders as
______________
84Ignatieff, Virtual War, p. 106.
85This is not to suggest that one should draw any particular comfort from the appar-
ent fact that NATO’s failure to take out more than a token number of VJ tanks was
largely irrelevant to the overall outcome of Allied Force. For one thing, had NATO
been able to render the VJ’s Kosovo corps ineffective during the air war’s initial month,
Milosevic may well have capitulated earlier, to the relief of both NATO and the Koso-
var Albanians. Second, and more important, the mission of finding, identifying, and
destroying dispersed and concealed enemy tanks is not going to go away, and the U.S.
Air Force will likely be asked again in some future contingency to attack fielded enemy
forces under comparably challenging circumstances. Civilians in senior leadership
positions who recall the more optimistic early claims on behalf of the air war’s accom-
plishments in this respect will naturally expect air power to perform effectively. For-
tunately, despite charges from some that the Air Force sought to play down its difficul-
ties in this regard in the early aftermath of Allied Force, its leadership has frankly
owned up to those difficulties and has initiated measures aimed at improving its ca-
pability. I am grateful to my RAND colleague Bruce Pirnie for directing my attention
to this point.
it was of any inherent deficiencies in the air weapon. By ruling out

before the fact even a ground threat, let alone any serious prospect of
an early ground invasion, the Clinton administration and NATO en-
sured that air power would be stressed to the fullest when it came to
attempts to engage fielded enemy forces.
STRAY WEAPONS AND THE LOSS OF INNOCENTS

Pressures to avoid civilian casualties and unintended damage to
nonmilitary structures were greater in Allied Force than in any previ-
ous campaign involving U.S. forces. Nevertheless, despite rules of
engagement characterized by USAF Major General Charles Wald as
being “as strict as I’ve seen in my 27 years in the military,” there were
more than 30 reported instances throughout the air war of unin-
tended damage caused by errant NATO munitions or mistakes in tar-
geting, including a dozen highly publicized incidents in which civil-
ians were accidentally killed.86 The first serious loss of civilian lives
occurred on April 12, when an electro-optically guided AGM-130 re-
leased by an F-15E struck a targeted rail bridge over the Jusna
Morava river in Kosovo on the Belgrade-Skopje line 300 km southeast
of Belgrade just as a passenger train full of noncombatants, in a
tragic moment of fateful timing, happened to be crossing it.87 Bel-
grade later reported that more than 55 civilians had been killed in
that incident. Two days later, in the worst case of collateral damage
______________
86Joel Havemann, “Convoy Deaths May Undermine Moral Authority,” Los Angeles
87Indeed, the train entered the AGM-130’s field of regard so close to the moment of
weapon impact that the F-15E weapon systems officer (WSO) who was controlling the
guiding weapon noted that he had not even seen it until the videotape of his cockpit
display was played back during the subsequent mission debriefing. As a measure of
the extent to which F-15E aircrews, like all others, were disciplined to honor the
strictest collateral-damage avoidance rules, there were numerous instances in which
the WSO dragged the selected impact point of a guiding AGM-130 off the designated
aim point to an open area at the last moment because the target looked through the
weapon’s EO seeker head like a house or some other potential opportunity for collat-
eral damage. In a similar illustration of such discipline, one videotape of an AGM-130
attack on an enemy fuel storage tank as the weapon neared impact showed the tar-
geted tank to be empty while others around it were full. Nevertheless, despite the
WSO’s natural temptation, the guiding weapon was not slewed at the last moment to-
ward a more lucrative target because the empty fuel tank happened to be the one to
which the approved DMPI had been assigned. Conversation with USAF F-15E air-
crews, 492nd Fighter Squadron, RAF Lakenheath, England, April 27, 2001.
to have occurred at any time throughout the operation, attacks

against presumed enemy military vehicles at two sites in southwest-
ern Kosovo near the town of Djakovica were said to have killed nu-
merous ethnic Albanian refugees when USAF F-16 pilots mistook
civilian vehicles for a convoy.88
These and similar possible target identification errors resulted, in at

least a few instances, from constraints imposed by the requirement
that NATO aircrews remain above 15,000 ft to avoid the most lethal
enemy infrared SAM and AAA threat envelopes, which made visual
discrimination between military and civilian traffic difficult at best.
Discriminate attacks against moving military vehicles amid a virtual
sea of civilian refugees typically bordered on being an impossible
mission when pilots orbiting at medium altitudes could not deter-
mine for sure whether a convoy consisted of military trucks, military
vehicles carrying refugees, or civilian vehicles. General Wald, the
deputy director of strategic planning on the Joint Staff and
the commander at Aviano during Operation Deliberate Force in
1995, conceded that “the job is about as hard as it’s going to get for
targeting.”89
Another contributing factor was the occasional tendency of allied

aircrews to maneuver their aircraft in such a way as to put clouds
within the targeting pod’s field of view between the aircraft and the
target, thus blocking the laser beam illuminating the target and de-
priving the weapon of guidance. On April 6, near the end of the sec-
ond week, the first LGB went astray in that manner, hitting an
apartment building in the small town of Aleksinac 100 miles south-
east of Belgrade and reportedly killing at least seven civilians and in-
juring dozens more. The intended target had been an artillery
brigade headquarters, but the bomb’s steering toward its desired
mean point of impact was disrupted by clouds that deflected the
laser beam after weapon release.
In the case of the Djakovica incident noted above, there were initial
reports that Yugoslav aircraft had intentionally attacked the civilian
______________
88Rowan Scarborough, “As Strikes Mount, So Do Errors,” Washington Times, May 11,
1999.
89 Robert Wall, “NATO Shifts Tactics to Attack Ground Forces,” Aviation Week and
Space Technology, April 12, 1999, p. 23.
tractors and wagons near Prizren. Those reports ultimately proved

groundless, although Pentagon officials did confirm that the Yu-
goslav air force was still operating low-flying Galeb ground-attack
jets and attack helicopters.90 In all events, the alleged occurrence of
an inadvertent bombing attack on noncombatant civilians took place
at midday, despite the greatest operational discipline on the part of
the involved USAF pilots. The F-16 strike force leader, who was op-
erating as an airborne forward air controller (FAC-A), determined the
initial convoy to be made up of uniformly sized, colored, and spaced
military vehicles whose occupants seemed engaged in systematic
house-burning. Extensive radio discussion then ensued between the
FAC-A and the ABCCC stressing the need to avoid inadvertently
harming any Kosovar refugees. The ABCCC, backstopped by an or-
biting UAV, confirmed the convoy to be a valid military target and
marshaled as many fighters against it as were available in the im-
mediate target area.
During the course of the precision attacks with 500-lb GBU-12 LGBs
that then ensued, it was reported as “possible” that some of the ve-
hicles may have been civilian tractors, at which point the FAC-A im-
mediately called all fighters off “high and dry” (clear of the target
area with their armament switches deselected), and the ABCCC, in
turn, requested reverification of the targets as hostile. At that point,
nearby OA-10s were called in so that their pilots might reconnoiter
the situation and provide such reverification with onboard nine-
power space-stabilized binoculars. One OA-10 pilot reported observ-
ing definite military vehicles but also multicolored and possibly
civilian vehicles, whereupon the FAC-A terminated all further at-
tacks. Afterward, Serb news reports claimed that 80 civilians had
been killed, although the persistent ambiguities were such that
NATO only conceded that it “may have attacked” civilian vehicles.
Some reports suggested that the civilians involved had been
machine-gunned rather than bombed, and eyewitnesses on the
ground reported the use of human shields in the convoys and nearby
Serb mortar fire at the same time the convoy was being attacked by
the F-16s. The commander of the 31st Air Expeditionary Wing whose
F-16s were involved in the tragedy, Brigadier General Leaf, later told
______________
90Michael Dobbs and Karl Vick, “Air Strikes Kill Scores of Refugees,” Washington Post,
April 15, 1999.
reporters that the incident involved “a very complicated scenario,

and we will never be able to establish all the details.” He further
stated that he could not explain the bodies of the civilians that had
been shown on Serbian television and conceded only, in light of the
ambiguous evidence, that there “may have been” unintended civilian
fatalities.91
The extraordinary media attention that was given to events like these
attested to what can happen when incurring zero noncombatant ca-
sualties becomes not just the goal of strategy but also the expecta-
tion. Thanks to unrealistic efforts to treat the normal friction of war
as avoidable human error, every occurrence of unintended collateral
damage became overinflated as front-page news and treated as a
blemish on air power’s presumed ability to be consistently precise.
Indeed, the added constraints imposed on NATO aircrews as a result
of such occasional tragic occurrences indicated the degree to which
modern air power has become a victim of its own success. During
the Gulf War, cockpit video images of LGBs homing with seemingly
unerring accuracy down the air shafts of enemy bunkers were spell-
binding to most observers. Yet because of that same seemingly
unerring accuracy, such performance has since come to be expected
by both political leaders and the public alike. Once zero collateral
damage becomes accepted as a measure of strategy success, not only
air power but all forms of force employment get set up to be judged
by all but unreachably high standards. Inevitably, any collateral
damage then caused during the course of a campaign becomes grist
for domestic critics and the enemy’s propaganda mill. Anthony
Cordesman rightly noted how characterizations of modern precision
bombing as “surgical” overlook the fact that patients still die on the
operating table from time to time.92 Nevertheless, a nontrivial num-
ber of proposed sorties in Operation Allied Force were either can-
celed outright or aborted at the last minute before any weapons were
released because their targets (wryly characterized by some USAFE
staffers as “morally hardened”) could not be positively identified or
______________
91Videotaped press statement by Brigadier General Daniel Leaf, USAF, Brussels, Bel-
gium, NATO Office of Information and Press, April 19, 1999.
92Anthony H. Cordesman, “The Lessons and Non-Lessons of the Air and Missile War
in Kosovo,” unpublished draft, Center for Strategic and International Studies, Wash-
ington, D.C., July 20, 1999.
because of the perceived risk of causing collateral damage. At best,

that made for a necessarily constrained and therefore inescapably
inefficient air operation compared to the standard set earlier in
Desert Storm.
A bevy of criticism arose from some quarters after the bombing

ended alleging that many of the 500 or more Yugoslav and Kosovar
Albanian civilians who lost their lives to collateral damage incidents
had died needlessly as a direct result of NATO attack aircraft having
been kept above 15,000 ft in the interest of minimizing the likelihood
of losing friendly lives. Critics further charged that operating at that
altitude had somehow been risk-free, cowardly, and even immoral
on the part of NATO’s aircrews.93 In league with other detractors of
the way the air war was conducted, strategist Edward Luttwak, for
example, characterized 15,000 ft as a “not-optimal” but “ultra-safe”
altitude from which allied pilots might carry out “perfectly safe
bombing.”94
In point of fact, 15,000–20,000 ft was precisely the “optimal” altitude

block from which to conduct LGB attacks—not only to keep the at-
tacking aircraft clear of short-range air defenses in the immediate
target area but, more important, to give the LGB time to acquire the
target and assume a stabilized glide. Contrary to the suggestions of
critics, operating at medium altitude provides no protection what-
ever against radar-guided SAMs. It merely puts attacking aircraft
outside the lethal envelope of “trash fire” threats (small arms, AAA,
and infrared SAMs). These threats are impossible to detect in a
timely way and offer little or no warning of imminent danger; as a re-
sult, they cannot be countered very effectively. Indeed, operating at
medium altitude actually increases the risk of being engaged by un-
negated enemy radar-guided SAMs because the aircraft can no
longer take advantage of terrain-masking opportunities. The more
important point, however, is that when medium-altitude attack tac-
______________
93Typical of such baseless charges was the reference by one pundit to the “low alti-
tudes at which tactical attacks work,” yet where “pilots risk getting killed” (William
Pfaff, “After NATO’s Lies About Kosovo, It’s Time to Come Clean,” International Her-
ald Tribune, May 11, 2000) and the allegation by another that “avoiding risk to pilots
multiplied the risk to civilians exponentially” (James Carroll, “The Truth About
NATO’s Air War,” Boston Globe, June 20, 2000).
94Edward N. Luttwak, “Give War a Chance,” Foreign Affairs, July/August 1999, p. 40.
tics are employed, the timeline for target acquisition and weapons
guidance is substantially longer, thus improving the chance of
achieving a hit.
Even assuming the absence of undetectable “trash fire” threats, it is

by no means a foregone conclusion that had allied aircrews routinely
descended to lower altitudes in an effort to identify their targets
more positively, the incidence of unintended collateral damage oc-
currences would have been that much lower. To begin with, VJ and
MUP troops were highly accomplished at camouflage and hiding,
and they made frequent use of the civilian populace as human
shields. Moreover, in Kosovo, where most of the inadvertent civilian
fatalities occurred, the mandated altitude floor was not invariably
15,000 ft as the critics implied. On the contrary, once operations
against dispersed and hidden VJ forces in Kosovo began in earnest in
mid-April, FACs were cleared down to 5,000 ft as necessary to make
positive target identifications, and strike aircraft could descend to as
low as 8,000 ft for a nonprecision dive-bomb delivery.
Even at those lower altitudes, however, positive identifications

tended to be difficult, although in one case, as noted above, USAF
OA-10 pilots using nine-power space-stabilized binoculars managed
to observe civilians intermingled with a VJ truck convoy after one
vehicle had already been hit, as a result of which the ongoing attack
was instantly terminated.95 As a rule, however, routinely going lower
and accepting the increased risk of losing an aircraft in the hope of
doing better target discrimination would not, in all likelihood, have
produced the desired result. True enough, flying even as low as 100
ft above ground level might have enabled NATO pilots to distinguish
civilian from military traffic in a few fleeting moments, if that traffic
happened to lie almost directly underneath the aircraft’s flight path.
Yet operating that close to the ground at normal fighter airspeeds
(500 nautical miles per hour or more) in defended airspace would
have offered zero perspective and zero precision-attack capability. It
also would have increased the chance of NATO aircraft losses to
enemy “trash fire” and just possibly brought about the overall failure
______________
95“NATO Jets May Have Erred in Convoy Attack, General Says,” Aerospace Daily, April
20, 1999, p. 102.
rather than success of Allied Force as a result. Moreover, hidden

targets would still have remained hidden.
The point of the foregoing is that for the kinds of circumstances that
repeatedly occasioned the accidental loss of civilian life in Allied
Force, the United States, to say nothing of its NATO allies, has yet to
develop fail-safe target discrimination capabilities and tactics for use
either above or below 15,000 ft. As a result, it has had little choice but
to rely on draconian rules of engagement (ROE), which are designed
to hedge on the side of caution yet are anything but foolproof. In one
case during Allied Force in which the ROE worked as intended, a
USAF pilot was directed not to attack a confirmed SA-6 launcher be-
cause it was parked immediately adjacent to a civilian structure in a
village. There were other reported instances in which precision
munitions in the process of guiding were deliberately steered away
from targets at the last minute to avoid harming civilians who had
not been seen in the target area until after weapon release.96 In the
most egregious instance in which the ROE regime appears to have
failed, however, namely, the tragedy involving the convoy along the
Djakovica road in Kosovo, the FAC who was coordinating the attack
had been given a positive identification by the ABCCC that was
completely consistent with the prevailing ROE. Upon observing that
the vehicles were uniformly colored and evenly spaced, the FAC
declared the convoy to be a valid target. He had also been given
ABCCC approval to clear the fighters under his control to drop at will
after one F-16 orbiting overhead had drawn fire from one of the con-
voy’s vehicles.97
______________
96John A. Tirpak, “The State of Precision Engagement,” Air Force Magazine, March
2000, p. 26.
97 It further bears stressing in this regard that most cases of unintended damage
resulting in civilian deaths occurred inside targeted buildings, which were
prespecified in the ATO and against which NATO aircrews were not free to exercise
real-time discretion. Other such cases were occasioned by munitions failures such as
faulty cluster-bomb fuses or laser target designators that were disrupted by smoke or
clouds while a weapon was guiding. Neither had anything to do with weapon-release
altitude. The only clear case of noncombatant fatalities that can be even indirectly
ascribed to altitude was the April 14 Djakovica convoy incident, during which the
attack was immediately called off once the target identification error was discovered.
The solution to such challenges lies not in more relaxed operating

restrictions but rather in the development of better tactics, tech-
niques, procedures, and equipment—perhaps beginning with a more
aggressive and effective use of offboard platforms like UAVs to per-
form combat identification and to provide cueing for engaged shoot-
ers.98 Unfortunately, the sensor-to-shooter links that have been re-
fined to a high art over the years for the air-to-air arena, such as the
E-3 AWACS and the joint tactical information distribution system
(JTIDS) carried by some F-15s, remain far less developed for ground-
attack operations when it comes to situation characterization and
target identification.99 In the absence of such capabilities, flying
lower in Allied Force not only would not have solved the target
identification problem, it would also have rendered weapons deliver-
ies less accurate and, as a result, probably compounded the collateral
damage problem rather than ameliorating it. As matters stood, al-
though regrettable tragedies did occur because of occasional mis-
directed weapons, the munitions and tactics used by NATO in
Operation Allied Force made the air effort a record-setter when it
came to achieving its declared goals with a minimum of collateral
damage for an operation of that magnitude. Indeed, given the high
volume of ordnance that was expended over the course of the 78-day
air war, it is most remarkable—even astonishing—that the incidence
of unintended civilian fatalities was not higher.
______________
98Email from Lieutenant Colonel James Tubbs, AF/XPXQ, to Colonel James Callard,
AF/XPXS, February 11, 2000. Lieutenant Colonel Tubbs was the operations officer of
the 510th Fighter Squadron flying F-16CGs out of Aviano Air Base during Operation
Allied Force.
99Although, as in Desert Storm, AWACS generally provided a superb threat picture to
allied pilots operating in hostile airspace, at least one specific instance of friction was
reported by a USAF F-15C pilot who downed a Yugoslav MiG-29 during a day defen-
sive counterair mission on March 26. The pilot complained that the supporting
AWACS controller “did not have any inkling [that] someone was flying on the other
side of the border, although he was real good at calling out every friendly west of us”
(email communication to the author, June 4, 1999). The F-15 pilot further charged
that the supporting AWACS was still unaware of the MiG-29’s presence even after ini-
tial moves had commenced. The intercepting pilot accordingly assessed the assumed
threat aircraft to be hostile by origin, since there were no NATO offensive counterair
missions airborne at the time. Only after the engagement was fully joined and the
F-15 pilot had visually confirmed his target to be a MiG-29 did the AWACS controller
finally report two possible hostile contacts in lead-trail formation.
THE CHINESE EMBASSY BOMBING

By far the most consequential instance of unintended bomb damage
in Allied Force occurred on May 7, when three JDAMs intended for
the headquarters of a Yugoslav arms agency were dropped instead
with unerring accuracy by a B-2 on the Chinese embassy in Belgrade.
That colossal blunder was reminiscent of the ill-fated attack on the Al
Firdos bunker by an F-117 during Desert Storm, which accidentally
killed more than a hundred Iraqi women and children who, unbe-
known to U.S. target planners, had been sleeping inside in the false
belief that it offered them shelter. The inadvertent bombing of the
Chinese embassy, which killed four occupants who happened to be
in the targeted portion and sent 26 more embassy staffers to the hos-
pital, became the latest of more than a dozen strikes in Allied Force
that had gone awry by that time. Not only did the bombing cause a
huge international uproar, it dramatized yet again how seemingly
“tactical” errors can have immensely disproportionate strategic con-
sequences. Among other things, the event triggered a diplomatic
crisis of the first order between Washington and Beijing, disrupted
moves to negotiate an end to the Kosovo conflict, and prompted a
politically directed halt to any further bombing of targets in Belgrade
for two weeks thereafter.100
At least two failures seem to have accounted for the inadvertent

bombing. First, CIA officials who nominated the intended target
wrongly deduced where it was located in Belgrade. Second, those
same officials were unaware that the actual targeted building was the
Chinese embassy, which had been moved there from another site
four years before. During Desert Storm, target planners almost al-
ways had knowledge of all off-limits buildings in and around Bagh-
dad, including foreign embassies, and they put red circles around
those buildings on planning maps to ensure that they would not be
inadvertently struck. Gulf War planners were also more proactive in
updating the no-strike list, to include having U.S. officials contact
______________
100 The error was also reminiscent of earlier damage to the French embassy in Tripoli,
Libya, in 1986 during the joint U.S. Air Force–U.S. Navy Operation El Dorado Canyon
against Libya’s ruler, Moammar Khaddafi, caused when the bomb fragmentation pat-
tern from a preceding F-111 forced the trailing pilot to shift course, inadvertently
sending his bombs into the embassy. That, however, was an operational error occa-
sioned by the heat of battle, not a planning error committed by target nominators.
foreign governments directly whenever there was any doubt about

the location of their embassies.101 In this case, although the target
development process most definitely included the creation and con-
tinual updating of a “no-strike” list of facilities, locations, and as-
sorted other entities that was duly vetted throughout the intelligence
community, U.S. officials admitted afterward that they had relied
on an outdated map of Belgrade. Some laid the blame on a budget-
cutting decision by the Clinton administration in 1996 to fold the
CIA’s National Photographic Interpretation Center (NPIC) into the
Defense Department’s National Imagery and Mapping Agency
(NIMA), which had prompted many of NPIC’s most experienced ana-
lysts to quit in protest.
In the immediate aftermath of the blunder, Secretary Cohen said:

“Clearly, faulty information led to a mistake in the initial targeting of
this facility. In addition, the extensive process in place used to select
and validate targets did not correct this original error.” 102 Cohen
added that the bombing had resulted not from a mechanical or hu-
man mistake but from “an institutional error.” 103 It was later de-
termined that the error had occurred in considerable part because of
the intense pressure that was being applied at the time by General
Clark for planners to come up with 2,000 suggested targets in Yu-
goslavia, prompted by the scramble for targets that had commenced
once the air war’s first few disappointing nights made it clear that
Milosevic was not about to fold quickly as had originally been hoped.
It was in this forced atmosphere of trying to find and justify 2,000
plausible targets at any cost that the CIA’s Counter-Proliferation Di-
vision, which had no particular expertise with respect either to Yu-
goslavia or to targeting, was led to submit the CIA’s first target nomi-
nation in Allied Force.
As it turned out, U.S. intelligence had the correct street address for
the intended target, which was a Yugoslav weapons-producing
agency called the Federal Directorate of Supply and Procurement.
______________
101 David A. Fulghum and Robert Wall, “Intel Mistakes Trigger Chinese Embassy
Bombing,” Aviation Week and Space Technology, May 17, 1999, p. 55.
102 Eric Schmitt, “Aim, Not Arms, at the Root of Mistaken Strike on Embassy,” New
York Times, May 10, 1999.
103 Paul Richter and Doyle McManus, “Pentagon to Tighten Targeting Procedures,”
Los Angeles Times, May 11, 1999.
Yet when overhead imagery was examined to match up the address

with the intended target, responsible individuals at CIA selected the
wrong building. The actual target turned out to have been on the
same street, only a block away to the south. The map used had been
created in 1992 and updated in 1997. It did not, however, show the
Chinese embassy at its current location, to which it had moved in
1996. No one in the planning loop had thought to check the match-
up of the target address with its presumed location, because no one
had any reason to believe that there might be a problem in the mak-
ing. One midlevel CIA analyst who was familiar with the intended
target reportedly “was concerned, raised some questions, and they
did not get resolved.” Doubts about the target’s validity also were
aired at the working level at USEUCOM, but those concerns were
never passed up to more senior levels before the strike.104 Afterward,
in a classic case of closing the barn door after the horse had escaped,
NATO officials cited a new “iron-clad requirement” that targets be
reviewed by people who had first-hand knowledge of them.105 Yet
despite that belated measure, on the first day after NATO’s bombing
of Belgrade resumed two weeks later, attacking aircraft inadvertently
damaged the residences of the Swedish, Spanish, and Norwegian
ambassadors, the Libyan embassy, and a hospital in which four
civilians were killed.106
Perhaps predictably when viewed in hindsight, more than a few peo-

ple around the world came to conclude in the early wake of the Chi-
nese embassy bombing that notwithstanding the U.S. government’s
insistent claims to the contrary, the bombing had, in fact, been not
only far from accidental, but planned with calculated intent from the
very start. Much of the apparent strength of this conspiracy theory
stemmed from the fact that the three JDAMs that were dropped by
the B-2 had, all too conveniently, landed squarely on that part of the
embassy that housed the office of the defense attaché and the em-
bassy’s intelligence cell, the latter of which was widely believed in
______________
104 Vernon Loeb and Steven Mufson, “CIA Analyst Raised Alert on China’s Embassy,”
105 Eric Schmitt, “Pentagon Admits Its Maps of Belgrade Are Out of Date,” New York
Times, May 11, 1999, and Bradley Graham, “U.S. Analysts Misread, Relied on Outdated
Maps,” Washington Post, May 11, 1999.
106 Steven Pearlstein, “NATO Bomb Said to Hit Belgrade Hospital,” Washington Post,
May 21, 1999.
informed circles to be the single largest Chinese collection center in

all of Europe.107
One can readily understand how that curious coincidence might

have helped energize Chinese allegations, which continue to this
day, that the bombing of the embassy was intentional. Yet as much
as one might wish to savor the thought that U.S. planners may have
been just clever enough to contrive to take out a Chinese SIGINT site
that was suspected of providing aid and comfort to the enemy while
maintaining a reliable basis for plausible denial, it defies credibility
to believe that those responsible for implementing Allied Force, at
whatever level such putative machinations may have occurred, at-
tacked the offending part of the embassy with premeditation. Al-
though truth is indeed stranger than fiction on occasion, no coalition
of democratic partners—least of all one led by an official Washington
that, since Watergate, has become famously reputed for leaking like a
sieve at even the slightest hint of high-level impropriety—could have
pulled off such a stratagem without it being exposed. Ivo Daalder
and Michael O’Hanlon perhaps best clinched this point when they
wrote that the strongest proof of the groundlessness of the conspir-
acy theory was that “the attack’s predictable damage—not only to
U.S.-PRC relations but even to NATO solidarity—was far too great to
justify the military benefit of silencing any Chinese military or intelli-
gence assistance to Serbia that could theoretically have been pro-
vided from that building.”108
TASK FORCE HAWK

As noted earlier in Chapter Three, within days after Operation Allied
Force commenced, General Clark asked the Army to deploy a contin-
gent of its AH-64 Apache attack helicopters to the combat zone to
provide a better close-in capability against enemy tanks and APCs
than that offered by fixed-wing fighters, which remained restricted to
operating at medium altitudes. Clark initially had hoped to deploy
this force to Macedonia, where the roads and airfields were better
______________
107 Steven Lee Myers, “Chinese Embassy Bombing: A Wide Net of Blame,” New York
108 Ivo H. Daalder and Michael E. O’Hanlon, Winning Ugly: NATO’s War to Save
Kosovo, Washington, D.C., The Brookings Institution, 2000, p. 147.
and the terrain less challenging. The Macedonian government, how-

ever, declined to grant permission because it was already swamped
by the flood of Kosovar refugees, so Clark sought Albania instead as
the best available alternative. 109 Within four hours, NATO had ap-
proved Clark’s request. It took more than a week, however, for the
U.S. and Albanian governments to endorse the deployment. That
approval finally came on Day 12 of Allied Force. The U.S. Defense
Department at first indicated that it would take up to 10 days to de-
ploy the package. In the end, it took 17 days just to field the first
battalion of Apaches, which arrived in Albania on April 21.
At first glance, the idea of using Apaches to reinforce NATO’s fixed-

wing aircraft seemed entirely appropriate, considering that the AH-
64 had been acquired by the Army expressly to engage and destroy
enemy armor. As Pentagon spokesman Kenneth Bacon put it in an-
nouncing the deployment, they would offer NATO “the type of tank-
killing capability that the bad weather has denied us . . . the capabil-
ity to get up close and personal to the [VJ] units in Kosovo.”110 In a
normal weapons load, the Apache mounts up to 16 Hellfire antitank
missiles, 76 folding-fin antipersonnel rockets, and 1,200 rounds of
30mm armor-piercing ammunition. With that armament, it gained
deserved distinction by destroying more than 500 Iraqi armored ve-
hicles during Operation Desert Storm. In Desert Storm, the Apaches
had deployed as an organic component of two fully fielded U.S. Army
corps. But in this case, the Army was being asked by SACEUR to
cobble together an ad hoc task force designed to operate essentially
on its own, without the backstopping support of a fielded U.S.
ground combat presence in the theater. The Army is not configured
to undertake such ad-hoc deployments, and its units do not train for
them. Instead, an Apache battalion normally deploys only as part of
a larger Army division or corps, with all of the latter’s organically
attached elements.
______________
109 Another reported problem with the Macedonia basing option was the fact that it
would have been a violation of the Dayton accords to station any offensive forces
within the territorial confines of the former Yugoslavia. Albania was thus the only real-
istic alternative.
110 Bradley Graham and Dana Priest, “Allies to Begin Flying Refugees Abroad,” Wash-
ington Post, April 5, 1999.
Accordingly, the Army was driven by its own standard operating pro-
cedures to supplement the two Apache battalions with an additional
heavy contingent of ground forces, air defenses, military engineers,
and headquarters overhead. As the core of this larger force comple-
ment, now designated Task Force (TF) Hawk, the Apaches were
drawn from the Army’s 11th Aviation Brigade stationed at Illesheim,
Germany. The deployment package included, however, not only the
two battalions of AH-64s, but also 26 UH-60L Blackhawk and
CH-47D Chinook helicopters from the 12th Aviation Regiment at
Wiesbaden, Germany. Additional assets whose deployment was
deemed essential for supporting the Apaches included a light in-
fantry company; a multiple-launch rocket system (MLRS) platoon
with three MLRS vehicles; a high-mobility multipurpose wheeled
vehicle (HMMWV, or “humvee”) antitank company equipped with 38
armed utility vehicles; a military intelligence platoon; a military po-
lice platoon; and a combat service support team. The Army further
determined a need for its Apaches to be accompanied by a mecha-
nized infantry company equipped with 14 Bradley AFVs; an armor
company with 15 M1A2 Abrams main battle tanks; a howitzer battery
with eight 155mm artillery pieces; a construction engineer company;
a short-range air defense battery with eight more Bradley AFVs
armed with Stinger infrared SAMs; a smoke generator platoon; a
brigade headquarters complement; and diverse other elements. In
all, to backstop the deployment of 24 attack helicopters to Albania,
TF Hawk ended up being accompanied by a support train of no fewer
than 5,350 Army personnel.
To be sure, there was a legitimate force-protection rationale behind

this accompanying train of equipment and personnel. Unlike the
Marines, who deployed 24 F/A-18D fighters to Hungary only a few
weeks thereafter and had them flying combat missions within days
with nothing even approaching TF Hawk’s overhead and support
baggage, Army planners had to be concerned about the inherent
risks of deploying a comparable number of Apaches on terrain that
was not that of a NATO ally, that lacked any semblance of a friendly
ground force presence, and that could easily have invited a VJ cross-
border attack in the absence of a U.S. ground force sufficient to ren-

der an attack an unacceptable gamble for VJ commanders.111
As one might have expected with so much additional equipment and

personnel, however, the Apache deployment soon encountered the
predictable consequences of the Army’s decision. It was at first esti-
mated that 200 USAF C-17 transport sorties would be needed to air-
lift the assorted support elements with which the Apaches had been
burdened. (The Tirana airport lacked the required taxiway and ramp
specifications to accommodate the more capacious C-5.) In the end,
it took more than 500 C-17 sorties, moving some 22,000 short tons in
all, to transfer TF Hawk in its entirety to Albania. Commenting later
on the deployment, one Army officer complained that the Army is
“still organized to fight in the Fulda Gap.” Even the outgoing Army
chief of staff, General Dennis Reimer, admitted in an internal memo
to senior Army staff officers once the deployment package had finally
been assembled in theater that the manifold problems encountered
by TF Hawk had underscored a “need for more adaptive force pack-
aging methodology.”112
In all events, the Apaches with their attached equipment and per-
sonnel arrived in Albania in late April. No sooner had the Army de-
clared all but one of the aircraft ready for combat on April 26 when,
only hours later, one crashed at the Tirana airfield in full view of re-
porters who had been authorized to televise the flight. (The 24th
Apache had developed hydraulic trouble en route and remained on
the ground in Italy.) Neither crewmember was injured, but the acci-
dent was an inauspicious start for the widely touted deployment.
Less than two weeks later, on May 5, a second accident occurred, this
time killing both crewmembers during a night training mission some
46 miles north of Tirana. The aircraft was carrying a full load of
______________
111 That said, it bears noting that the threat of Serbian forces coming across the Alba-
nian border did not appear to be a matter of great concern to anyone in the Allied
Force command hierarchy before the arrival of TF Hawk, even though there were U.S.
troops already on the ground in Albania as a part of JTF Shining Hope, the Albanian
refugee relief effort, who were not provided with any comparable force-protection
package.
112 Elaine M. Grossman, “Army’s Cold War Orientation Slowed Apache Deployment to
Balkans,” Inside the Pentagon, May 6, 1999, p. 6. Notably, the C-17 demonstrated for
the first time the ability to air-deliver a significant Army force of M1 tanks, M2 AFVs,
MLRSs, howitzers, and engineering equipment.
weapons and extra fuel. A subsequent investigation concluded that

the first accident had been caused by the pilot’s having mistakenly
landed short of his intended touchdown point.113 The second was
attributed to an apparent failure of the tail rotor because the aircraft
had been observed to enter a rapid uncontrolled spiral during the
last moments before its impact with the ground.
As of May 31, the cost of the TF Hawk deployment had reached $254
million, much of that constituting the expense for the hundreds of
C-17 sorties that had been needed to haul all the equipment from
Germany to Albania, plus the additional costs of building base camps
and port services and conducting mission rehearsals.114 Yet despite
SACEUR’s intentions to the contrary, the Apaches flew not a single
combat mission during the entire remainder of Operation Allied
Force. The reason given afterward by the JCS chairman, General
Shelton, was that Serb air defenses in Kosovo, although noticeably
degraded by early May, remained effective enough to warrant keep-
ing the Apaches out of action until SEAD operations had “reduced
the risk to the very minimum.”115
In a final coda to the Army’s plagued TF Hawk experience, Shelton

conceded in later testimony to the Senate Armed Services Committee
that “the anticipated benefit of employing the Apaches against dis-
persed forces in a high-threat environment did not outweigh the risk
to our pilots.”116 Shelton added that by the time the deployment had
reached the point where the Apaches were ready to engage in com-
bat, VJ ground formations were no longer massed but had become
dispersed and well hidden. Moreover, he went on to note, the
weather had improved, enabling Air Force A-10s and other fixed-
wing aircraft to hunt down dispersed and hidden enemy forces while
______________
113 Paul Richter and Lisa Getter, “Mechanical Error, Pilot Error Led to Apache
Crashes,” Los Angeles Times, May 13, 1999.
114 Ron Lorenzo, “Apache Deployment Has Cost Quarter Billion So Far,” Defense
Week, June 7, 1999, p. 6.
115 Molly Moore and Bradley Graham, “NATO Plans for Peace, Not Ground Invasion,”
116 Sheila Foote, “Shelton: Risk Was the Key in Decision Not to Use Apaches,” Defense
Daily, September 10, 1999, p. 2.
incurring less risk from enemy infrared SAMs, AAA, and small-arms
fire than the Apaches would have faced.117
Beyond the problems created for the deployment by the Army’s deci-
sion to bring along so much additional overhead, there was a break-
down in joint doctrine for the combat use of the helicopters that was
disturbingly evocative of the earlier competition for ownership and
control of coalition air assets that had continually poisoned the rela-
tionship between the joint force air component commander (JFACC)
and the Army’s corps commanders during Desert Storm.118 The is-
sue stemmed in this case from the fact that the Army has tradition-
ally regarded its attack helicopters not as part of a larger air power
equation with a theater-wide focus, but rather as an organic maneu-
ver element fielded to help support the ground maneuver needs of a
division or corps. Apache crews typically rely on their own ground
units to select and designate their targets. Yet in the case of Allied
Force, with no Army ground combat presence in theater to speak of,
they would either have had to self-designate their targets or else rely
on Air Force forward air controllers flying at higher altitudes to des-
ignate for them. The idea of using Apaches as a strike asset in this
manner independently of U.S. ground forces was simply not recog-
nized by prevailing Army doctrine. On the contrary, as prescribed in
Army Field Manual FM 1-112, Attack Helicopter Operations, an AH-64
battalion “never fights alone. . . . Attacks may be conducted out of
physical contact with other friendly forces,” but they must be
“synchronized with their scheme of maneuver.” FM 1-112 expressly
characterizes deep-attack missions of the sort envisaged by Clark as
“high-risk, high-payoff operations that must be exercised with the
utmost care.”119
______________
117 True enough, the terrain and weather presented by Kosovo were more challenging
than the open and featureless Iraqi desert, where the Apaches had performed so ef-
fectively against enemy armor in Desert Storm. Yet the biggest concern in the minds
of many U.S. leaders was the specter of a replay of the 1993 “Bloody Sunday” horror in
Mogadishu, Somalia, with dead Army Rangers and crewmembers from downed Black-
hawk helicopters being dragged through the streets on live television worldwide.
118 David Atkinson and Hunter Keeter, “Apache Role in Kosovo Illustrates Cracks in
Joint Doctrine,” Defense Daily, May 26, 1999, p. 6.
119 Quoted in Elaine M. Grossman, “As Apaches Near Combat, White House Seeks
Diplomatic Solution,” Inside the Pentagon, May 6, 1999, p. 7.
In light of this, the Army’s V Corps commander, Lieutenant General

John Hendrix, was willing to have the Apaches included in the
USEUCOM Air Tasking Order (ATO), but demurred on having them
incorporated as well into the separate NATO ATO, notwithstanding
General Short’s insistence that such inclusion would be essential in
any situation in which the attack helicopters were ever committed to
actual combat. Apart from that, however, Short never sought opera-
tional control of the Apaches or attempted to task them. He also of-
fered to provide TF Hawk as much operational support (including
EA-6B jamming support) as possible, and even went so far as to pro-
pose to subordinate himself and his CAOC to Hendrix, who as V
Corps commander was also the ultimate commander of TF Hawk, as
a supporting (as opposed to supported) combat element.120
In the end, an agreement was reached that included the Apaches

with all other ATO missions yet left to Hendrix’s discretion much es-
sential detail on mission timing and tactics. A window was provided
in the ATO such that the Apaches would be time-deconflicted from
friendly bombs falling from above and also assured of some fixed-
wing air support. However, the agreement reached in the end was so
vague that it allowed each service to claim that it maintained tactical
control over the Apaches in the event they were ever committed to
combat. For their part, Army officers insisted that fire support for the
AH-64s would come only from MLRS and Army tactical missile sys-
tems (ATACMS) positioned on the Albanian side of the border. That
doctrinal stance was enough all by itself to ensure that the Apaches
would never see combat, considering that the massive MLRS and
ATACMS fires envisaged for any AH-64 operations would have rained
literally multiple thousands of CBU submunitions all over Kosovo in
an indiscriminate attempt to suppress enemy AAA and IR SAMs, a
tactic that was out of the question from the very start, given NATO’s
determination to avoid any significant incidence of noncombatant
casualties. In contrast, Air Force planners maintained that excluding
the Apaches from CAOC control would increase their level of risk by
depriving them of support from such key battlespace awareness as-
sets as Joint STARS, Rivet Joint, Compass Call, and the EA-6B. As a
USAF officer attached to Hendrix’s deep operations coordination cell
______________
120 Telephone conversation with Lieutenant General Michael Short, USAF (Ret.),
August 22, 2001.
(DOCC) reportedly put it, “they do not know, nor do they want to
know, the detailed integration required to get the Prowler to jam the
priority threats, provide acquisition jamming on the correct azimuth,
etc. . . . The benefits of integrating with platforms like Compass Call,
Rivet Joint and others are off their radar scope.”121
After Allied Force ended, the assistant chief of staff for operations at
Supreme Headquarters Allied Power Europe (SHAPE), USAF Major
General John Dallager, touched the heart of the overriding interests
and equities at stake here when he stated, during a briefing at a
NATO Reaction Force Air Staff conference on JFACC issues: “Clearly
the JFACC’s authority must not infringe upon operational C2
[command and control] relationships within and between national
or service commands and other functional commands. But to
ensure deconfliction of simultaneous missions and to minimize the
risk of fratricide, all air operations within the [joint operating arena]
must be closely coordinated by the JFACC through the ATO . . .
process. This last point may be difficult to swallow for land and
maritime commanders, but if air history teaches us anything, it is
that air, the truly joint activity, needs to be coordinated centrally if
we are to make efficient use of scarce resources and if we are to avoid
blue-on-blue.”122
______________
121 Elaine M. Grossman, “Army Commander in Albania Resists Joint Control over
Apache Missions,” Inside the Pentagon, May 20, 1999, p. 9. In his memoirs, Clark later
scored this article for “personally attacking Jay Hendrix and claiming, among other ac-
cusations, that he would not allow the Apache sorties to appear on Short’s Air Tasking
Order.” Clark made no attempt to refute that accusation, however, but merely dis-
missed it as the complaint of a “disgruntled Air Force officer” whose “misunder-
standing, communicated without perspective to friends in other units, suddenly
surfaced to make news weeks after it had been written, after the problems it ad-
dressed, if real then, had been corrected.” General Wesley K. Clark, Waging Modern
War: Bosnia, Kosovo, and the Future of Combat, New York, Public Affairs, 2001, p. 320.
122 Major General John Dallager, USAF, “NATO JFACC Doctrine,” briefing at a confer-
ence on “The NATO Joint Force Air Component Commander Concept in Light of the
Kosovo Air Campaign,” Headquarters NATO Reaction Force Air Staff, Kalkar, Ger-
many, December 1–3, 1999. It might be noted in passing here that another Army–Air
Force difference of view that had an even greater operational impact than the joint
doctrinal disagreement discussed above (because all involved had to live through its
consequences) was the disconnect between the two services at Tirana as to who was in
charge of the airfield and force protection, a disconnect that, according to one senior
USAF planner who was involved, created “some real problems.” Comments on an
earlier draft by Brigadier General Robert Bishop, Hq USAF/XOO, April 17, 2001.
Interestingly, the Army leadership in the Pentagon seemed far more

disposed than General Hendrix, at least in principle, to assign opera-
tional control of the Apaches to the CAOC. The incoming Army vice
chief of staff, Lieutenant General Jack Keane, frankly commented at
an industry symposium that “it boggles my mind, but we still have
senior leaders, people who wear stars . . . that don’t recognize that if
you’re going to fly Apaches at a distance and range, it’s got to be on
the [air tasking order].” General Keane added that the Apaches had
to be under the operational control of the JFACC in the Army’s “self-
interest” because that arrangement offered a more effective way of
employing them in this particular instance: “The JFACC should de-
termine what the Apache’s targets are as a result of the entire re-
sponsibility he has in conducting that air campaign.” He further
noted that the JFACC had the comparative advantage of being able to
retask combat assets based on real-time intelligence, something that
the Army could take advantage of as well if it could get itself out of
“the business of being myopic about ground operations.” In closing,
he acknowledged that in the Army, “we’ve got this nagging fear that
somehow, if we turn over our organization to somebody in another
uniform, that that organization is somehow going to suffer as a result
of that. And I just fundamentally disagree with that.” 123
In yet further testimony to the ill-fated nature of the Army’s TF Hawk

experience, an internal Army memorandum written after Allied
Force ended acknowledged that the aircrews sent with the Apaches
had been both undertrained and underequipped for their intended
mission. In a report to the incoming chief of staff, General Eric Shin-
seki, then–Brigadier General Richard Cody, the Army’s director of
operations, resources, and mobilization, warned that because of
those shortcomings, “we are placing them and their unit at risk when
we have to ramp up for a real world crisis.” Cody, who earlier had
planned and executed the Army’s highly successful Apache opera-
tions during the Gulf War, noted that more than 65 of the assigned
aviators in TF Hawk had less than 500 hours of flight experience in
the Apache and that none were qualified to fly missions requiring
night-vision goggles. He further noted that the radios in the de-
ployed Apaches had insufficient range for conducting deep opera-
tions and that the crews were, in the absence of night-vision goggles,
______________
123 Ibid.
dependent solely on their forward-looking infrared (FLIR) sensors.

Given the rugged terrain, unpredictable weather, and poorly marked
power lines that crisscrossed Kosovo, relying on FLIR alone, he sug-
gested, “was not a good option.” Moreover, he added, in order for
the Apaches to have flown the required distances and crossed the
high mountains of Kosovo, Hellfire missiles would have had to be
removed from one of their two wing mounts to free up a station for
auxiliary fuel tanks. As for the MANPADS threat, Cody remarked that
“the current suite of ASE [aircraft survivability equipment] was not
reliable enough and sometimes ineffective.”124
The TF Hawk experience underscored how little the U.S. Army, by its
own leadership’s candid admission, had done since Desert Storm to
increase its capacity to get to an emergent theater of operations
rapidly and with sufficient forces to offer a credible combat presence.
Shortly after the Gulf War, the Army’s leadership for a time enter-
tained the thought of reorganizing the service so that it might be-
come more agile by abandoning its structure of 10 combat divisions
and opting instead for 25 “mobile combat groups” of around 5,000
troops each. Ultimately, however, the Army backed away from that
proposed reform, doing itself out of any ability to deploy a strong
armored force rapidly and retaining the unpalatable alternatives of
airlifting several thousand lightly armed infantrymen to a theater of
conflict within days or shipping a contingent of 70-ton M1A2 Abrams
main battle tanks over the course of several months.125
On his second day in office as the Army’s new chief of staff, General
Shinseki acknowledged that the Army had been poorly prepared to
move its Apaches and support overhead to Albania. Part of the
problem, he noted fairly, was that the only available deployment site
that made any operational sense had poor rail connections, a shallow
port, and a limited airfield capacity that could not accommodate the
Air Force’s C-5 heavy airlifter. However, he admitted that the Army
______________
124 George C. Wilson, “Memo Says Apaches, Pilots Were Not Ready,” European Stars
and Stripes, June 20, 1999.
125 Thomas E. Ricks, “Why the U.S. Army Is Ill-Equipped to Move Troops Quickly into
Kosovo,” Wall Street Journal, April 16, 1999. The most fully developed and widely
cited articulation of this proposed Army reorganization, which failed to take root, may
be found in Colonel Douglas A. Macgregor, USA, Breaking the Phalanx: A New Design
for Landpower in the 21st Century, Westport, Connecticut, Praeger, 1997.
was nevertheless overdue to develop and act on a plan to make its

heavy forces more mobile and its lighter forces more lethal.126 In
what may have presaged a major shift in Army force development
policy for the years ahead, he declared: “Our heavy forces are too
heavy and our light forces lack staying power. Heavy forces must be
more strategically deployable and more agile with a smaller logistical
footprint, and light forces must be more lethal, survivable, and tacti-
cally mobile. Achieving this paradigm will require innovative think-
ing about structure, modernization efforts, and spending.”127
One positive role played by TF Hawk after the KLA’s counteroffensive

began registering effects in late May was the service provided by the
former’s counterbattery radars in helping NATO fixed-wing pilots
pinpoint and deliver munitions against enemy artillery positions. Its
TPQ-36 and TPQ–37 firefinder radars were positioned atop the hills
adjacent to Tirana to spot Serb artillery fire and backtrack the air-
borne shells to their point of origin. Army EH-60 helicopters and
RC-12 Guardrail electronic intelligence aircraft were further able to
establish the location of VJ command posts whenever the latter
transmitted. Although TF Hawk’s Apaches and other combat assets
never saw action, its ISR assets exerted a significant influence on the
air effort at one of its most crucial moments. The KLA’s counterof-
fensive had forced the VJ to mass its forces and maneuver, to com-
municate by radio, and to fire artillery and mortars to protect itself.
In response, the sensors of TF Hawk, operating in conjunction with
the Army’s Hunter UAVs, spotted VJ targets and passed that infor-
mation on to those in the command loop who could bring air-
delivered ordnance to bear in a timely manner. “The result,” said a
retired Army three-star general, “was that NATO air power was finally
able to target precisely and hit the Serb army in the field. The Koso-
vars acted as the anvil and TF Hawk as the eyes and ears of the black-
smith so that the hammer of air power could be effective.”128
Echoing this conclusion, USAFE’s commander, General Jumper, con-
______________
126 Eric Schmitt, “New Army Chief Seeks More Agility and Power,” New York Times,
June 24, 1999.
127 “Shinseki Hints at Restructuring, Aggressive Changes for the Army,” Inside the
Army, June 28, 1999, p. 1.
128 Lieutenant General Theodore G. Stroup, Jr., USA (Ret.), “Task Force Hawk: Beyond
Expectations,” Army Magazine, August 1999.
firmed that the counterbattery radars of TF Hawk had played “a very

big part” in allied targeting during the final stages of Allied Force.129
Another bright spot in the otherwise troubled TF Hawk experience

was the USAF air mobility system’s superb performance in opening
up the Rinas air base in Albania and flowing forces and relief supplies
into it. The combined efforts of USAFE’s Air Mobility Operations
Command Center (AMOCC), the Allied Force Air Mobility Division,
USAFE’s 86th Contingency Response Group at Ramstein Air Base,
Germany, and multiple supporting Air Mobility Command entities
resulted in a stand-out success amid the generally dismal story of TF
Hawk’s immobility and the Army’s persistent go-it-alone approach
to command relations and putting the Apaches into the ATO. Simply
put, the C-17 made the TF Hawk movement possible. (See Figure 6.3
for the sharp spike in C-17-delivered short tonnage connected with
TF Hawk from the second week of April through the first week of
May.) No other aircraft could have done the job—yet another
testimonial to the direct-delivery concept that shaped the aircraft’s
design and got it through one of the most hard-fought acquisition
battles in the USAF’s history. Thanks to the C-17 acquisition, TF
Hawk (despite its near-fatal growing pains) got in, and many
thousand Albanian refugees survived—two signal accomplishments
of what the commander of the U.S. Army in Europe, General Mont-
gomery Meigs, later called one of the most successful airlift opera-
tions in history.130
SHORTCOMINGS IN INTELLIGENCE CYCLE TIME

Commanders and other air operators throughout the course of Allied
Force found themselves repeatedly frustrated by the amount of time
it often took to cycle critical information about enemy pop-up
______________
129 Response to a question at an Air Force Association Eaker Institute colloquy,
“Operation Allied Force: Strategy, Execution, Implications,” held at the Ronald
Reagan International Trade Center, Washington, D.C., August 16, 1999.
130 Comments on an earlier draft by Brigadier General Robert Bishop, Hq USAF/XOO,
April 12, 2001, and Colonel Robert Owen, Hq AMC, May 10, 2001. See also General
Charles T. Robertson, Jr., USAF, commander in chief, U.S. Transportation Command,
and commander, Air Mobility Command, “Air War Over Serbia: A Mobility
Perspective,” briefing charts, 2000, Hq USAFE/SA library.
RAND MR1365-6.3
1600
C-17
1400
C-130
C-141
1200
C-5
Short tons delivered
1000 Other
800
600
400
200
0
24
31
14
21
28
12
19
26
9
ril
ay
ne
ne
ch
ch
ril
ril
ril
ay
ay
ay
Ap
Ju
Ju
Ap
Ap
Ap
M
ar
ar
M
Figure 6.3—Short Tonnage Delivered by USAF Airlift
targets of opportunity from sensors to shooters who were positioned

to engage them effectively. Although the requisite architecture was
in place throughout most of the air war once a flexible targeting cell
had been established by the end of the first month, it lacked a
sufficiently high-volume data link with enough channels to quickly
get the information where it needed to go.
To be sure, there were occasional instances of major success stories.

For instance, the U-2 demonstrated its ability to be retasked in real
time to image a reported SA-6 site, data-link the resulting imagery via
satellite back to its home base at Beale AFB, California, within min-
utes for an assessment of the target’s coordinates, and have the re-
sults transmitted back to the cockpit of an F-15E just as its pilot was
turning inbound toward the target to fire an AGM-130.131 In another

such case, on Day 4 a Navy TLAM on short notice successfully at-
tacked a “target of opportunity” believed to have been a pop-up
MiG-29 detected on the runway at Batajnica by real-time imagery
from a U-2.132 Although those examples were not representative,
they previewed the sort of fusion toward which the U.S. ISR system is
heading and represented what USAF Lieutenant General Marvin Es-
mond later described as “the first-ever distributed ISR architec-
ture.”133
More typically, however, target images from Predator UAVs flying

over Kosovo would be transmitted to the CAOC in real time, only to
encounter difficulty being forwarded from there to operating units in
time for them to be tactically useful. In addition, the Joint STARS
crew complement was found to be too small to accommodate many
of the data processing and reporting demands it was asked to handle.
The aircraft was said to require either more battle managers inte-
grated closely enough into the commander’s loop for targets to be
identified and attacked in near-real time, or wider-band data links to
ground stations, where a larger number of mission specialists could
do the analysis and handling.134
Yet a third bottleneck identified was the classified worldwide Inter-

net link called SIPRNET (Secure Internet Protocol Router NETwork),
upon which USEUCOM’s Joint Analysis Center (JAC) at RAF
Molesworth, England, relied heavily. As a rule, intelligence sources
would forward proposed target materials to Molesworth for valida-
tion, with the JAC staff tasking additional intelligence collection as
deemed necessary. That process would have been all but impossible
______________
131 The AGM-130 is a rocket-boosted variant of the electro-optical and infrared guided
GBU-15 2,000-lb PGM featuring midcourse GPS guidance updates. At the start of the
air war, 200 of these weapons had been fielded, and those used were pulled from Air
Combat Command’s Weapons System Evaluation Program (WSEP), leaving no muni-
tions for training. William M. Arkin, “Kosovo Report Short on Weapons Performance
Details,” Defense Daily, February 10, 2000, p. 2.
132 Ibid.
133 Lieutenant General Marvin R. Esmond, testimony to the Military Procurement
Subcommittee, House Armed Services Committee, Washington, D.C., October 19,
1999.
134 David A. Fulghum, “Lessons Learned May Be Flawed,” Aviation Week and Space
Technology, June 14, 1999, p. 205.
without the aid of the Internet, which made for vastly more rapid
worldwide information availability than did the former hard-copy
practices. Frequently, however, because of the absence of institu-
tionalized procedures, the use of SIPRNET made for confusion and
difficulty in finding some target materials on short notice. In addi-
tion, real-time target information would be withheld from U.S. allies
as U.S. officials argued over who should be allowed to see what. Fi-
nally, NIMA was frequently slow to deliver overhead photography of
proposed targets and of targets already attacked, which in turn
slowed the battle-damage assessment process and the decision as to
whether to retarget a previously attacked site. One informed source
commented that ISR fusion worked better in Allied Force than it did
during Desert Storm, but that it still rated, at best, only a grade of
C-plus in light of what remained to be done. In contrast, what gener-
ally worked well was the “reach-back” procedure first pioneered in
Desert Storm, in which commanders and planners in the forward
theater used secure communications lines to tap into information
sources in the intelligence community in Washington and else-
where. 135
AIRSPACE AND TRAFFIC FLOW MANAGEMENT

A major concern for Allied Force mission planners entailed the co-
ordination of air operations with so many allied aircraft transiting the
relatively dense and compact airspace between Italy and the Balkans.
Among other things, the CAOC coordinated operations by some 200
NATO tanker aircraft operating out of eight countries to support
strikers flying from 15 bases in Germany, France, Italy, Hungary,
Spain, the United Kingdom, and the United States.136 There were
numerous reported instances of near-midair collisions caused by
marginal weather and an insufficiency of battle management infor-
mation relayed by AWACS to friendly aircraft operating in and near
the combat zone. Mission planners at the CAOC sought to deconflict
allied aircraft by parceling out the most impacted airspace so that
only a given number of friendly aircraft would be operating inside
______________
135 Rowan Scarborough, “Kosovo Target Data Stalled in Transit,” Washington Times,
July 28, 1999.
136 Tim Ripley, “Tanker Operations,” World Air Power Journal, Winter 1999/2000,
p. 121.
any block at a given time. The danger of midair collisions was of

particular concern in designated engagement zones, or “kill boxes,”
in the KEZ, with only a few allied aircraft being permitted to operate
within a given box at any time for that reason. Both the E-3 AWACS
and the EC-130 ABCCC carried copies of the daily ATO, which al-
lowed them to keep track of scheduled flight operations and remind
allied aircrews of pertinent details as necessary. Another problem
caused by the unusually congested airspace over and near Yugoslavia
entailed linking some combat aircraft with their assigned tankers,
particularly the German Tornado ECR variants and the EA-6B, which
lacked air-to-air radars and had to be vectored to their tankers by
AWACS.137
In an important contribution to easing the air traffic nightmare that

threatened to ensue over the Adriatic and in the adjacent airspace as
the air effort unfolded, Italian air traffic authorities lent their exper-
tise to the CAOC’s air traffic control cell in order to make key staffers
there more familiar with Italian airspace structure and regulations.
They also dispatched a representative to the military cell of the re-
gional civilian air traffic control (ATC) center to smooth out potential
difficulties in controlling the heavy flow of ATO sorties going in and
out of the area of responsibility (AOR). Measures taken to manage
that flow and to deconflict it from civil traffic included closing the
airspace over parts of the Adriatic, establishing a no-fly zone encom-
passing the airports of Bari and Brindisi, suppressing all or parts of
some airways, establishing a special corridor to permit the transit of
Italian airspace by air traffic entering from outside the AOR, provid-
ing a system of safe operating routes to allow the departure and re-
turn of combat aircraft loaded with weapons operating from Italian
air bases, and establishing six emergency weapons jettison areas in
international waters and six active inflight refueling zones over the
Adriatic 24 hours a day.
Not surprisingly, the Italian ATC system experienced considerable

difficulty in handling this large volume of daily traffic. To begin with,
because of the air war’s length and the shortage of available con-
trollers, ATC found it a major challenge to maintain round-the-clock
control of all the active and alternate military airfields that were in-
______________
137 Wall, “Airspace Control Challenges Allies.”
volved in air operations. Second, Eurocontrol experienced problems

managing civil aviation flight plans, given the density of military traf-
fic, and was not always able to maintain the impermeability of the
posted no-fly zone over the Adriatic. Third, ATC was frequently un-
able to track military aircraft operating from the several aircraft car-
riers that were deployed in the Adriatic and, for that reason, faced
serious deconfliction problems with civil traffic flying along the
southern air routes toward Greece and Turkey. Fourth, communica-
tion problems were often encountered between and among the vari-
ous agencies engaged in air traffic flow management, such as airfield
control towers, approach and departure control centers, military re-
gional control, air defense radars, and AWACS. Finally, there was far
too little time available to debug, test, and properly validate these
highly jury-rigged arrangements. Although the system worked in the
end with no catastrophic or otherwise untoward incidents, numer-
ous aircrews reported that the aerial traffic jams of ingressing and
egressing NATO aircraft transiting the AOR throughout Allied Force
often appeared more dangerous than the threat presented by Ser-
bia’s SAMs and AAA. 138
As it unfolded and expanded in scope and intensity, Operation Allied

Force became the largest civilian emergency ever confronted by the
airlines, although it produced little major traffic dislocation in the
end. Before the cold war ended, there had been only two options
from which to choose—either a peacetime operating mode, with the
military taking only a small portion of the available airspace and time
for training, or a wartime mode, with no civil operations whatever
and unrestricted military flying. This time, as NATO’s top official on
civil airspace put it, the coalition was “waging what we may plainly
call war in a localized area of Europe, while throughout the rest of the
continent it was business as usual.”139 The situation required air
traffic controllers to reroute as many as 8,000 airliners a day on some
occasions. One concern was that inconveniencing civilians at peak
summer travel time would erode public support and cause a back-
______________
138 Colonel E. Baldazzi, Italian Air Force, “Host Nation Support for the Kosovo Air
Campaign,” briefing at a conference on “The NATO Joint Force Air Component Com-
mander Concept in Light of the Kosovo Air Campaign,” Headquarters NATO Reaction
Force Air Staff, Kalkar, Germany, December 1–3, 1999.
139 Joseph Fitchett, “For NATO, Keeping Peak Air Traffic on the Go Was a Critical
Goal,” International Herald Tribune, March 31, 2000.
lash against the effort. Another was to avoid any replay of the
downing of an Iranian airliner, which the cruiser USS Vincennes
mistook for an Iranian F-14 over the Persian Gulf in 1988. That latter
concern led to a double-checking of identification procedures for
electronically identifying aircraft operating in and near the combat
zone. Toward the end of the air war, NATO finally succeeded in eas-
ing the airspace congestion problem at least marginally, when it in
effect opened a second front by initiating Marine F/A-18D opera-
tions out of Hungary and USAF fighter operations out of Turkey.
DEFICIENCIES WITH RESPECT TO SPACE

Fortunately, U.S. space superiority was not challenged during Op-
eration Allied Force. Against the remote yet distinct possibility that
Milosevic and his erstwhile supporters in Moscow might somehow
have sought to do so, however, the enemy’s space order of battle,
rudimentary though it may have been, was never seriously exam-
ined. Nor was the vulnerability of U.S. space systems sufficiently as-
sessed and hedged against using the needed countermeasures. Other
space-related problems were also highlighted by the Allied Force ex-
perience. With respect to ISR, intelligence collectors and combat air-
crews both had repeated difficulty finding mobile targets. Adverse
weather and enemy camouflage, concealment, and deception mea-
sures presented additional complications, with the result that the
“kill chain” was too long by a discomfiting margin. Relatedly, space-
based weather support suffered. For example, there was no continu-
ous weather coverage of the theater of operations, so some sched-
uled strike missions may have been needlessly canceled because
available weather information was not current. Battlespace charac-
terization also suffered because of a lack of enough space-derived
imagery of the right kinds.140
As for other deficiencies in the availability of on-orbit assets, some

satellites that had been tasked primarily against the Middle East and
Pacific basin were recommitted to the Balkans, leaving important ar-
______________
140 “Space Support to Operation Allied Force: Preliminary Lessons Learned,” briefing
to the author by Colonel Robert Bivins, Director of Operations, U.S. Air Force Space
eas of interest uncovered in other theaters.141 Moreover, the United

States was shown to continue to lack a real-time targeting capability
and to suffer significant problems with respect to real-time distribu-
tion, all of which pointed to the still-unresolved challenge of getting
the right information to the right people at the right time. To be sure,
information cycle time was compressed significantly in comparison
to earlier aerospace operations, as attested by one case in which a
single TLAM was targeted and launched early during Allied Force
against a MiG-29 that had suddenly been detected in the open at a
Serbian air base. However, there was no mechanism available for
providing shooters in near-real time with radar imagery and other
intelligence gathered by the multitude of collection platforms and
surveillance systems that were available and functioning. Joint
STARS is slated to receive an upgrade that will permit it to transmit a
map through a satellite uplink directly into a fighter cockpit, but that
capability is not yet in place.142 Also, U.S. space-based intelligence
assets, including NRO’s classified ELINT and SIGINT satellites, DSCS,
and other systems, were shown not to have improved greatly since
Desert Storm. As one U.S. intelligence official noted, “three to four
hours is the best we can do” from target identification to weapons
delivery. 143 The good news in all this is that many needed fixes were
discovered to lie in the realm of essentially cost-free improvements
in techniques, tactics, and procedures rather than in more expensive
hardware solutions.
Finally, the Allied Force experience indicated that considerable room

remains for further progress in bringing operators to a fuller appre-
ciation of what space systems now have to offer. The director of
NRO, Keith Hall, commented after the air war ended that although
allied operators turned in an effective performance, they made some
important aspects of the operation harder for themselves than they
needed to be. Stressing that professional military education and offi-
cer specialization training at all levels in the four U.S. services still do
______________
141 Bill Gertz and Rowan Scarborough, “Dangerous Drawdown,” Washington Times,
April 30, 1999.
142 “Space Support to Operation Allied Force: Preliminary Lessons Learned,” briefing
to the author by Colonel Robert Bivins, director of operations, U.S. Air Force Space
143 Roy Bender, “Allies Still Lack Real-Time Retargeting,” Jane’s Defense Weekly, April
7, 1999.
not offer enough needed first-hand exposure to space systems and

their capabilities, he went on to say: “I impress upon [the service
chiefs] the need to organize, train, and equip to use this stuff if
they’re going to rely on it, and not just call up the NRO and say, ‘Can
you do this for us?’ when we’re engaged in an operation. . . . We’re
dealing with a situation where people are not trained, it hasn’t been
practiced in peacetime, and you have to scramble. . . . If they’re go-
ing to rely on it, they’re going to have to do their part of it.”144
Air Force space professionals would undoubtedly concur that short-

comings in the use of available space assets identified during Allied
Force highlighted a continuing need for more space involvement in
peacetime exercises, on the sensible premise followed as gospel for
years by fighter pilots that you should “train like you expect to fight.”
That means a need for advanced space education and training of a
highly specific and focused nature—not just greater “space aware-
ness”—for operators at all levels, from the most senior command
echelons all the way down to shooters working within tactical con-
fines. It also means a need for better development and documenta-
tion of space operational support capabilities and options in theater
contingency plans worldwide. Acknowledging this and more after
the air war was over, the commander in chief of U.S. Space Com-
mand, Air Force General Richard Myers, remarked frankly that in
terms of using space assets, the Kosovo operation was “probably the
best we’ve done—surely superior to Desert Storm from everything
we can learn. But there’s still a long way to go before space is really
integrated with the rest of the campaign.”145
INTEROPERABILITY PROBLEMS
One of the most surprising aspects of the Allied Force experience was
what it revealed about the extent of the discontinuity that had been
allowed to develop between U.S air power and that of most other
NATO allies who participated. 146 One concern had to do with inade-
______________
144 John Donnelly, “NRO Chief: Services Ill-Prepared to Work with Spy Satellites,”
Defense Week, July 12, 1999, p. 2.
145 Quoted in The Air War Over Serbia, p. 53.
146 For a fuller treatment of the allied contribution to the air war and the interoper-
ability problems that became manifest as a result of it, see John E. Peters, Stuart John-
quacies in the equipment operated by the allies. To begin with, there

was a pronounced dearth of interoperability with respect to rapid
and secure communications. Only at Aviano were there some old
STU-2 secure telephones that allowed the U.S. participants to trans-
fer classified information quickly to allied units. (The STU-3 secure
phone system used by U.S. forces was not available to the allies.)
Other classified communications required passing a hard copy of the
information by hand, repeating one of the worst command and con-
trol deficiencies that had been exposed earlier in Desert Storm, when
the ATO had to be flown every day to each of the Navy’s six partici-
pating aircraft carriers because the latter were not equipped to re-
ceive it electronically.
In addition, many NATO European fighters lacked Have Quick–type

frequency-hopping UHF radios and KY-58–like radios allowing en-
crypted communications. As a result, U.S. command and control
aircraft were often forced to make transmissions in the clear to those
fighters about targets and aircraft positions, enabling the enemy to
listen in and gain valuable tactical intelligence.147 Also, in at least
one case, British Harrier GR. Mk 7 pilots were said to have observed
suspected refugees in a convoy but were unable to communicate that
information to the ABCCC or to USAF F-16s operating in the same
area.148 For their part, U.S. aircraft equipped with JTIDS frequently
were not allowed to rely on that asset, but were instead obliged to use
voice communications to ensure adequate situation awareness for all
players, notably allied participants not equipped to receive JTIDS
signals. 149
_____________________________________________________________
son, Nora Bensahel, Timothy Liston, and Traci Williams, European Contributions to
Operation Allied Force: Implications for Transatlantic Cooperation, Santa Monica,
California, RAND, MR-1391-AF, 2001.
147 Since allied aircraft could not receive Have Quick radio transmissions and since
enemy forces made no effort to jam allied UHF communications, which Have Quick
was expressly developed to counter, the Have Quick capability was not used by U.S.
combat aircrews during Allied Force.
148 “NATO Jets May Have Erred in Convoy Attack, General Says,” p. 102.
149 David A. Fulghum and Robert Wall, “Data Link, EW Problems Pinpointed by Pen-
tagon,” Aviation Week and Space Technology, September 6, 1999, pp. 87–88. The JTIDS
offers aircrews a planform view of their tactical situation, as well as a capability for
real-time exchange of digital information between aircraft on relative positions,
weapons availability, and fuel states, among other things. It further shows the posi-
tion of all aircraft in a formation, as well as the location of enemy aircraft and ground
threats. Fighters can receive this information passively, without highlighting them-
Second, among all participating allied air forces, only U.S., British,
Canadian, French, Spanish, and Dutch combat aircraft had the abil-
ity to deliver LGBs without offboard target designation assistance.
General Short frankly admitted that he could not risk sending the air-
craft of many allied countries into harm’s way because of concern for
the safety of their pilots and for the civilian casualties that might be
caused by inaccurately aimed weapons. Largely for that reason,
around 80 percent of all strike sorties flown in Allied Force were car-
ried out by U.S. aircraft.
Additional problems making the job of AWACS operators difficult

included the absence of a robust alliance-wide IFF (identification
friend or foe) system, the lack of a capability to detect which SAM
systems were targeting allied aircraft, and the small number of non-
U.S. aircraft able to laser-designate targets, all of which inhibited the
usefulness of many allied assets. Some aspects of the discrepancy
between U.S. and allied capability were a result of the fact that the
European nations typically spend only half the annual U.S. percent-
age of defense expenditure on military procurement and a third of
the annual U.S. percentage on research and development.150 Others
merely reflected allied decisions to invest in different types of
equipment. Largely because of that asymmetry, however, the United
States provided almost all of the aerial intelligence employed in Al-
lied Force and selected virtually every target attacked in Operation
Allied Force. Commenting on these and other interoperability
problems, General Naumann expressed concern that the growing
technology gap between the United States and its allies could even-
tually lead to their inability to fight together or even communicate in
the same battlespace. 151
To be sure, not all participating allied air forces suffered equally pro-
nounced problems with respect to capability and versatility. The
Royal Netherlands Air Force, for example, not only kept its F-16s up
to date but also provided some aerial refueling capability. The Dutch
_____________________________________________________________
selves through radio voice communications. See William B. Scott, “JTIDS Provides
F-15Cs ‘God’s Eye View,’” Aviation Week and Space Technology, April 29, 1996, p. 63.
150 John D. Morrocco, “Kosovo Reveals NATO Interoperability Woes,” Aviation Week
and Space Technology, August 9, 1999, p. 32.
151 Barton Gellman and William Drozdiak, “Conflict Halts Momentum for Broader
Agenda,” Washington Post, June 6, 1999.
and the Belgians operated a total of 28 Block 15 F-16A/B midlife

update (MLU) aircraft as a single detachment at Amendola AB, Italy,
incorporating modifications that made the aircraft, to all intents and
purposes, Block 50-equivalents. A Dutch F-16 downed a Serb MiG-
29 with an AIM-120 AMRAAM during the opening night of the air
war, and another used its LANTIRN targeting pod to identify and
successfully attack a MiG-29 on the ground while ignoring several
decoys that were parked directly adjacent to it. According to the
principal Dutch airman assigned to the CAOC, the Royal Netherlands
Air Force (RNLAF) was “not 100-percent interoperable but close” and
was characterized by senior U.S. airmen as being “most definitely on
the A-Team.”152
Moreover, German and Italian Tornados contributed valuable SEAD

capabilities, firing some 37 percent of all HARM shots taken during
Allied Force. Seven of the nine allies contributing aircraft that
dropped bombs in the air war operated PGM-capable aircraft, which
at least made them effective in precision attacks in clear weather
against fixed targets. USAF Block 40 F-16CGs equipped with low-
altitude navigation and targeting infrared for night (LANTIRN)
targeting pods and using cooperative strike tactics designated targets
for numerous allied aircraft, including the Italian AMX, which were
capable of dropping LGBs but lacked any onboard self-designation
capability.
With the USAF now out of the manned tactical reconnaissance busi-
ness altogether and the Navy’s TARPS-equipped F-14s providing the
only remaining U.S. operational capability of that nature, three of the
five remaining French Mirage IVP supersonic bombers, since con-
verted to the reconnaissance role, added valuable support by being
flown daily when the weather permitted, accounting in the end for 20
percent of the Allied Force reconnaissance missions. Operating out
of Solenzara, Italy, they flew at 40,000–50,000 ft at a speed of Mach
2.05, typically entering the war zone over Belgrade and exiting over
Kosovo, covering some 20 targets on each flight in around 15 min-
utes. Returning traditional wet-film photographs to Solenzara, they
eventually developed a routine whereby high-quality images anno-
______________
152 Conversation with Major General P. J. M. Godderij, deputy commander in chief,
RNLAF, Scheveningen, the Netherlands, June 7, 2000.
tated with target information would be digitized for transmission to

the CAOC and to French headquarters in Paris.153
Finally, two decades of multinational training at Red Flag and else-

where paid off handsomely in Allied Force. There were no midair
collisions or other near-catastrophic aerial incidents resulting from
allies operating from their own private playbooks.
THE WAGES OF U.S. OVERCOMMITMENT

The demands placed by Allied Force on U.S. equipment and person-
nel underscored the extent to which the U.S. defense posture has
been stretched dangerously thin by the post–cold war force draw-
down and concurrent quadrupling of deployment commitments
worldwide. During the initial post–cold war decade of the 1990s, the
U.S. active-duty force in all services shrank by 800,000 personnel to
1.4 million, a reduction of more than one-third. The Army was cut
from 18 to 10 active divisions, the Navy diminished in size from 567
ships to just over 230, and the Air Force lost half of its 24 fighter
wings. Yet during that same period, the U.S. armed forces were
tasked with 48 major deployment missions overseas, in contrast with
only 15 between the time of the U.S. exit from Vietnam and the col-
lapse of the Soviet Union nearly two decades later.154
The first practical effect of this drawdown manifested during Allied

Force was the unexpectedly high rate at which scarce and expensive
consumables were being expended to meet the air war’s demands.
After only the first week, the Air Force found itself running low on
CALCMs, with the initial stock of 150 down to fewer than 100.155 The
Air Force had had preexisting plans in hand to convert 92 additional
nuclear-configured ALCMs to CALCMs, but that process was ex-
pected to take more than a year. JDAM was still being tested at the
time it was committed to combat. As of April 20, less than a month
into Allied Force, there were only 609 JDAM kits remaining in
______________
153 Chris Pocock, “Mirage IV Reconnaissance Missions,” World Air Power Journal,
Winter 1999/2000, p. 111.
154 Rowan Scarborough, “Record Deployments Take Toll on Military,” Washington
155 Hewson, “Operation Allied Force,” p. 21.
stock.156 The burdens placed by the air war’s demands on materiel

of all kinds prompted a rising groundswell of military complaints
that the results of seven years of underfunding were finally making
their impact fully felt.157
On that point, a memorandum from Air Combat Command (ACC) to

the Air Staff in late March frankly admitted that “our operational
units are suffering, with few serviceable engines [and] depleted
wartime spare kits.”158 ACC’s commander, General Hawley, re-
ported a month later that five weeks of bombing had left U.S. muni-
tions stocks, notably CALCM and JDAM, in critically short supply,
adding that “it’s going to be really touch-and-go as to whether we’ll
go Winchester [the pilot’s term for running out of ammunition] on
JDAMs.” Hawley warned that should a more serious crisis erupt
elsewhere, ACC would be “hard-pressed to give them everything that
they would probably ask for. There would be some compromises
made.”159 The later resort to an increased use of dumb bombs in Al-
lied Force was driven in part by the steady depletion of stocks of
precision munitions of all kinds.
Seeking an explanation for this increased stress on the U.S. defense

establishment across the board, General Hawley laid the blame
squarely on the nation’s military overcommitment: “I would argue
that we cannot continue to accumulate contingencies. At some
point you’ve got to figure out how to get out of something.” Hawley
added that because of a fourfold spike in the number of deployments
in the 1990s at the same time the force was undergoing a reduction
by half, “we are going to be in desperate need, in my command, of a
significant retrenchment in commitments for a significant period of
time. I think we have a real problem facing us three, four, five
______________
156 The principal deputy assistant secretary of the Air Force for acquisition at the time,
then–Lieutenant General Gregory Martin, acknowledged that the shortage of JDAMs
was the result of a conscious choice made five years ago to emphasize other procure-
ment needs. David A. Fulghum, “Bomb Shortage Was No Mistake,” Aviation Week and
Space Technology, May 17, 1999, p. 55.
157 See Rowan Scarborough, “Smaller U.S. Military Is Spread Thin,” Washington
158 Ibid.
159 Bradley Graham, “General Says U.S. Readiness Is Ailing,” Washington Post, April
30, 1999.
months down the road in the readiness of the stateside units.”160

Earlier during Allied Force, even before SACEUR’s twofold force in-
crease request was approved, Hawley cautioned that because of the
existing strain on the system, “if we deploy the additional forces that
are under consideration, those strains will become more evident,”
causing a “significant decline in the mission-capable rates” of the
remaining forces to as low as 50 percent or less for some aircraft
types.161
A second indication of the extent to which the U.S. military had come
to find itself strapped as a result of the force drawdown was the
sharply increased personnel tempo that was set in motion by the air
effort. In all, some 40 percent of the active-duty U.S. Air Force was
committed to Operation Allied Force and to the concurrent Opera-
tions Northern and Southern Watch over Iraq. That was roughly the
same percentage of Air Force personnel that had been committed
during Operation Desert Storm, when the total force was much
larger. Among other things, as noted earlier in Chapter Three, the
heightened personnel tempo obliged President Clinton to approve a
Presidential Selected Reserve Call-Up authorizing a summons of up
to 33,102 selected reservists to active duty. 162 It further prompted
the Air Force chief of staff, General Michael Ryan, to insist that the
USAF needed a recovery time no less than that routinely granted to
the Navy every time one of its carriers returns from a deployment.
Ryan flatly declared that “we are not a two-MTW [major theater war]
Air Force in a lot of areas, and one of them is airlift.” That shortfall
made for one of many reasons why the Air Force later insisted that it
needs 90 days to reconstitute its forces between MTWs.163
Earlier, as Allied Force entered its second month, Ryan told reporters
that “the U.S. Air Force is in a major theater war.” (He later amended
that remark to indicate that he had meant to say that the Air Force’s
commitment level included Operations Northern and Southern
______________
160 Ibid.
161 John A. Tirpak, “The First Six Weeks,” Air Force Magazine, June 1999, p. 27.
162 “U.S. Mobilizes Guard, Reserve for Balkan Duty,” Air Force Magazine, June 1999,
p. 16.
163 Vince Crawley, “Air Force Needs 90 Days Between Wars, Chief Says,” Defense Week,
August 9, 1999, p. 12.
Watch over Iraq.)164 In the eight years since Desert Storm, deploy-
ment demands on Air Force assets had never before exceeded the
level of two AEFs of around 175 aircraft each. NATO’s air war for
Kosovo, however, demanded four AEF-equivalents’ worth of USAF
assets. Then-acting Secretary of the Air Force F. Whitten Peters de-
clared that as a result, the AEF concept would need to be reexam-
ined.165
Third, the demands of Allied Force placed a severe strain on such low
density/high demand (LD/HD) aircraft as Joint STARS, AWACS, the
U-2, the B-2, the F-16CJ, and the EA-6B.166 So many of these scarce
assets were committed to the air effort that day-to-day training in
home units suffered major shortfalls as a result. The most acute
strains were felt in the areas of surveillance, SEAD, and combat
search and rescue. Almost every Block 50 F-16CJ in line service was
committed to support SEAD operations, necessitating a virtual halt
to mission employment training in the United States. (Figure 6.4
shows the overall USAF commitment to Allied Force, broken down
by aircraft type.)
Similarly, Vice Admiral Daniel Murphy, the commander of the 6th

Fleet, which provided the U.S. naval forces that were operating in the
Adriatic, reported that there was an insufficiency of EA-6B jammers
and they, along with their aircrews, were being worn out by the air
war’s demands.167 Almost half of the initial batch of 11 EA-6Bs used
to spearhead the air operation had been drawn from assets previ-
ously committed to Operation Northern Watch at Incirlik Air Base,
Turkey. Navy and Marine spokesmen declined to admit that their
EA-6Bs were being stressed to the danger point, but they did concede
that they were being run ragged trying to marshal enough aircraft out
______________
164 Tirpak, “The First Six Weeks,” p. 27.
165 John T. Correll, “Assumptions Fall in Kosovo,” Air Force Magazine, June 1999, p. 4.
166 Characteristics of LD/HD include single-unit asset, limited numbers of aircraft and
pilots, and likely tasking in more than one theater. Joint Vision 2010, the “revolution
in military affairs,” improved sensor to shooter links, and decisive attack operations all
depend on more support to LD/HD assets. They transcend individual service and
weapon system boundaries.
167 Dale Eisman, “Kosovo Lesson: Navy Says It Needs More High-Tech Tools,” Norfolk
Virginian-Pilot, June 10, 1999.
RAND MR1365-6.4
Total: 517
Intratheater
airlift A-10
(43) (40)
F-15C
Special operations/ (18)
rescue/other
(38)
F-15E
(32)
ISR*
(25)
F-16CG
(35)
KC-10
(24)
F-16CJ
(64)
KC-135
(151) F-117
(25)
B-52
B-2 B-1 (11)
(6) (5)
*ISR includes RQ-1, E-3, E-8, RC-135, U-2, and EC-130 ABCCC.
Figure 6.4—USAF Aircraft Types Employed
of the total inventory of 124 to support the launching of Allied

Force.168
An even greater demand was imposed on the Air Force’s various ISR
platforms, which left none available for day-to-day continuation
training once the needs of Allied Force were superimposed on preex-
isting commitments. During the time in question, the Air Force had
______________
168 Greg Seigle, “Prowler Jammers Used to Aid NATO Air Assault,” Jane’s Defense
Weekly, March 31, 1999.
only four operational E-8 Joint STARS aircraft, two of which were
committed to Allied Force (it has since acquired a fifth). As a result,
the Joint STARS community found itself so stripped of its most skilled
personnel that there was no instructor cadre left to work with new
crewmembers who were undergoing conversion training. The low
Joint STARS availability rate made for a typical Allied Force E-8 mis-
sion length of more than 17 hours, with the longest missions lasting
21 hours. It took two or more inflight refuelings and backup pilots
and crews to sustain each mission.169 Some Joint STARS aircraft
were flown at more than three times their normal use rates, creating
a major maintenance and depot backlog that would take months to
clear up. In all, U.S. LD/HD assets were stretched to their limit with
tasking demands whose reverberations will continue to be felt for
years in the areas of platforms, systems, reliability, parts, personnel,
retention, and replacement costs. On this point, Admiral Ellis cau-
tioned that the trend line is working in precisely the wrong direc-
tion—the demand for these assets in the future will only grow and
they should be viewed as national assets requiring joint funding, irre-
spective of service, as the highest priority. 170
Finally, Operation Allied Force exposed the extent to which U.S.

forces are being stretched to the limit to support real-world peace-
keeping and peacemaking commitments on a routine basis, while
also meeting the demands of engaging successfully in two simulta-
neous or near-successive major theater wars. In the prevailing de-
fense lexicon, Kosovo was supposed to be only a “smaller-scale con-
tingency.” Yet the number of U.S. aircraft committed to Allied Force
quickly approached the level of a major theater war and exposed
shortcomings in the availability of needed assets in all services. For
example, the diversion of the USS Theodore Roosevelt from the
Mediterranean to the Adriatic to support the air effort deprived U.S.
Central Command of a vital operational asset. Likewise, the later re-
deployment of the USS Kitty Hawk from the Pacific to the Persian
______________
169 Edmund L. Andrews, “Aboard Advanced Radar Flight, U.S. Watches Combat
Zone,” New York Times, June 14, 1999.
170 Briefing by Admiral James O. Ellis, USN, commander in chief, U.S. Naval Forces,
Anvil, “The View from the Top,” 1999.
Gulf deprived U.S. Pacific Command of a carrier in the western Pa-

cific for the first time since the end of World War II.
The Air Force was similarly forced to juggle scarce assets to handle
the overlapping demands imposed by Kosovo, Iraq, and Korea. It
positively scrambled to find enough tankers to support NATO mis-
sion needs in Allied Force. Ironically, both Kosovo and Iraq, in and
of themselves, represented lesser contingencies whose accommoda-
tion was not supposed to impede the U.S. military’s ability to handle
two major theater wars. Yet the burdens of both began to raise seri-
ous doubts as to whether the two-MTW construct, at least at its cur-
rent funding level, was realistic for U.S. needs. For example, when
USEUCOM redeployed 10 F-15s and 3 EA-6Bs from Incirlik to sup-
port Clark’s requirements for Allied Force, it was forced to suspend
its air patrols over northern Iraq immediately. Air patrols to enforce
the no-fly zone over southern Iraq were continued, but at a slower
operational tempo. The net result was U.S. aircraft being flown two
to three times more often than in normal peacetime operations.171
One example of the negative effects on combat readiness that

surfaced during Allied Force was the frequent and widespread com-
plaint by unit personnel in all services that their combat perfor-
mance suffered because their lack of prior training opportunities
with live weapons adversely affected their precision-weapons
employment techniques and procedures in actual combat. Indeed,
the majority of American bomb-droppers had never dropped a live
LGB in training. That shortfall in combat proficiency was partly a
reflection of limited range space, but it was also the result of under-
resourcing of combat units in the training-munitions category. Nu-
merous misses in Allied Force occurred because aircrews did not un-
derstand target-area effects such as thermal bloom, smoke, and dust,
which cannot be duplicated in peacetime training without live
weapon drops. By one informed account, civilians were injured in
Pristina and Surdulica as a direct result of smoke and IR bloom ef-
fects. Targets were also missed when aircrews discovered several
surprising effects in the LANTIRN system when using the combat
laser in the presence of clouds. The training laser (which is eye-safe)
______________
171 Elizabeth Becker, “Needed on Several Fronts, U.S. Jet Force Is Strained,” New York
fires at a much lower power and rate, with the result that the noted
effects were not discovered until they were actually seen in combat—
usually in the middle of a drop.172 Bowing to the inevitable, General
Shelton finally acknowledged the cumulative impact of these
multiple untoward trends when he admitted to Congress at the
beginning of May 1999 that there was “anecdotal and now mea-
surable evidence . . . that our current readiness is fraying and that the
long-term health of the total force is in jeopardy.”173
______________
172 Comments on an earlier draft by Hq USAF/XOXS, July 9, 1999.
173 Kate O’Beirne, “Defenseless: The Military’s Hollow Ring,” National Review, May 3,
1999, p. 18.
Chapter Seven
LAPSES IN STRATEGY AND IMPLEMENTATION
In the predictable rush to identify “lessons learned” that followed in

the wake of the air war’s successful outcome, senior administration
officials hastened to acclaim Operation Allied Force as “history’s
most successful air campaign.” 1 Yet NATO leaders on both sides of
the Atlantic had little to congratulate themselves about when it came
to the manner in which the air war was planned and carried out. On
the contrary, there was a dominant sense among both participants
and observers that the desultory onset of Allied Force and its later
slowness to register effects reflected some fundamental failures of
allied leadership and strategy choice.
Indeed, the six years that preceded Allied Force saw a clear regres-
sion in the use of air power after the latter’s casebook performance in
Desert Storm. With the singular exception of Operation Deliberate
Force in 1995, a trend toward what came to be called “cruise missile
diplomacy” had instead become the prevailing U.S. pattern, owing to
the ability of cruise missiles to deliver a punitive message without
risking the lives of any U.S. aircrews. The origins of this pattern went
back to June 1993, when President Clinton first ordered the firing of
several TLAMs in the dead of night against an empty governmental
building in Baghdad in symbolic reprisal for confirmed evidence that
Saddam Hussein had underwritten an assassination attempt against
former President George Bush.
______________
1Paul Richter, “U.S. Study of War on Yugoslavia Aimed at Boosting Performance,” Los
Angeles Times, July 10, 1999.
179
That trend was next reflected in the administration’s unwillingness

or inability to use air power decisively in dealing with Bosnian Serb
atrocities throughout the two years before Operation Deliberate
Force, and in the costly, yet apparently ineffectual, TLAM strikes
launched later by the administration against presumed assets of the
terrorist Osama bin Laden in Sudan and Afghanistan.2 It culminated
in the three-day Operation Desert Fox, a mini-air operation that was
executed against Iraq, to no significant consequence, at the very
height of President Clinton’s impeachment trial in December 1998.
Less than a year earlier, a more serious campaign plan called Opera-
tion Desert Thunder, set in motion shortly after Iraq had expelled the
UN’s arms inspectors in January 1998, was aborted by President
Clinton literally at the last minute, as allied strike aircraft were taxiing
for takeoff, in response to the extraction by UN Secretary General
Kofi Annan of an eleventh-hour, later unfulfilled, promise from Sad-
dam Hussein to permit UN inspections.3 In all of these cases, the
declared emphasis was merely on “degrading” or “damaging,” rather
than destroying, enemy assets, so that the operation could be termi-
nated at any moment in a manner allowing success to be declared.
That may have been the administration’s going-in hope for Opera-
tion Allied Force as well. Not long after the effort began, however,
senior U.S. military leaders began voicing off-the-record misgivings
over the slow pace of the air operation, its restricted target base, and
its rules of engagement that all but proscribed any serious applica-
tion of air power. One Air Force general spoke of officers in Europe
who had characterized the air war to date as “a disgrace,” adding that
“senior military officers think that the tempo is so disgustingly slow it
______________
2In fairness to the Clinton administration, it must be said that bombing the Bosnian
Serbs unilaterally was not a realistic option for the United States as long as three
NATO allies (France, the Netherlands, and the United Kingdom) had troops on the
ground who would have been helpless against Serb reprisals had U.S. air strikes taken
place. It was only after the UN Protection Force (UNPROFOR) pulled back into de-
fensible positions so that the Serbs could not take its troops hostage that Operation
Deliberate Force became politically feasible. Weakness on the ground can often
negate strength in the air.
3For an informed, if also sharply judgmental, account of this history, see Joshua Mu-
ravchik, “The Road to Kosovo, Commentary, June 1999, pp. 17–23. See also Lieutenant
Colonel Paul K. White, USAF, Crises After the Storm: An Appraisal of U.S. Air Opera-
tions in Iraq Since the Persian Gulf War, Military Research Papers No. 2, Washington,
D.C., Washington Institute for Near East Policy, 1999.
Lapses in Strategy and Implementation 181
makes us look inept.” 4 Another, harking back to the initial concept

of operations developed for Desert Storm, complained: “This is not
Instant Thunder, it’s more like Constant Drizzle.”5 Yet a third Air
Force general, reflecting the consensus of most airmen, commented
that “the hammer is working just fine. But when the blueprints have
to undergo revision each day by 19 separate architects before it is de-
termined where to drive the nail, one has to wonder what the final
product is going to look like.”6
Indeed, the highly politicized and sometimes seemingly random tar-

geting process was so cumbersome that Clark himself would discover
from time to time that he was stymied by the system as action time
neared. 7 The frequent hesitancy and indecision on the part of
NATO’s political leaders, and the resultant fits and starts which that
indecision inflicted on the daily target allocation machinery, ended
up producing what some uniformed critics later faulted as “ad hoc
targeting”: Air strikes were demanded on the same day that they had
been approved, missions that had not yet been approved were as-
signed to the JFACC, and those same missions were later removed
from the list at the last minute if they had not been approved by
NATO’s civilian authorities. The resulting confusion led the com-
mander in chief of Allied Forces in Southern Europe, Admiral James
Ellis, to complain: “We don’t like this kind of process where some-
thing could be left on [the ATO] by omission.”8 The burdensome
rules and restrictions that dominated the target approval process,
moreover, contributed to a defensive and reactive mind-set among
target planners and mission coordinators at the working level, who
______________
4 Rowan Scarborough, “U.S. Pilots Call NATO Targeting a ‘Disgrace,’” Washington
5 John D. Morrocco, David Fulghum, and Robert Wall, “Weather, Weapons Dearth
Slow NATO Strikes,” Aviation Week and Space Technology, April 5, 1999, p. 26.
6William M. Arkin, “Inside the Air Force, Officers Are Frustrated About the Air War,”
Washington Post, April 25, 1999.
7To illustrate, Clark recalled after the cease-fire that he would often have to call Solana
at the last minute with an urgent request like: “You’ve got to help me with target 183.
I need 183.” Michael Ignatieff, “The Virtual Commander: How NATO Invented a New
Kind of War,” The New Yorker, August 2, 1999, p. 34.
8Ibid.
were said by some to be locked into a resigned “we can’t do it” posi-
tion rather than amenable to a more creative “let’s try it” attitude.9
To be sure, it was not as though NATO’s uniformed professionals had

been railroaded into an operation against Milosevic without having
given it prior consideration. On the contrary, serious and detailed
options planning for an air operation of some sort against Yugoslavia
had begun at USAFE headquarters as far back as June 1998—plan-
ning that was never ultimately made use of for political reasons.
Nevertheless, it became clear, shortly after the bombing effort began,
that the relatively seamless performance by the coalition in Desert
Storm was not to be replicated in Allied Force. Instead, what un-
folded was a highly dissatisfying application of air power that showed
not only the predictable fits and starts of trying to prosecute a war
through an alliance of 19 members bound by a unanimity rule, but
also some failures even within the operation’s U.S. component to
make the most of what air power had to offer within the prevailing
constraints of alliance warfare.
ALLIED MISCALCULATIONS AND FALSE HOPES

To begin with, despite the ultimate success of Allied Force, a mis-
judgment of near-blunder proportions came close to saddling the
United States and NATO with a costly and embarrassing failure:
NATO’s leaders did not appreciate the historical and cultural impor-
tance of Kosovo to the Serbs and the consequent criticality of Kosovo
to Milosevic’s continued political livelihood. Fortunately for the al-
lies, their faulty assessment was not a show-stopper, although it eas-
ily could have been had Milosevic refrained from launching his eth-
nic cleansing campaign and instead merely hunkered down in a
defensive crouch to wait out the bombing in a contest of wills with
NATO. Once he elected to raise the stakes by proceeding with Op-
eration Horseshoe, however, NATO’s determination to prevail at all
cost deprived his strategy of any foundation it may previously have
had.
______________
9Roundtable discussion with Hq USAFE/XP, USAFE/DO, and USAFE/IN staff, Ram-
stein AB, Germany, May 2, 2001.
One reason for NATO’s overconfidence that air power alone would
suffice in forcing Milosevic to yield on Kosovo was almost surely a
misreading of the earlier Bosnian war and the role of Operation De-
liberate Force in producing the Dayton accords of 1995. As has been
widely noted since Allied Force ended, Bosnia was a part of the for-
mer Yugoslav Federation where Milosevic generally got what he
wanted and to which he was not particularly deeply attached. In the
negotiations that eventually yielded the Dayton accords, Milosevic
succeeded in keeping Kosovo unburdened by their strictures at the
price of abandoning Sarajevo to the Muslims, in a direct and outright
betrayal of his Bosnian Serb compatriots, because there was no sig-
nificant Serb minority living there.
In contrast, Kosovo was generally acknowledged to be of profound

historical importance for Serbia. Among other things, it contained
Kosovo Polje, the site where the Ottoman Turks defeated the Serb
kings in 1389. As journalist Michael Ignatieff has pointed out, “it was
here that the Kosovar lands passed under Turkish Ottoman control
for more than five centuries; it was here that the Serbian dream of re-
conquering Kosovo one day was born, a dream not realized until just
before World War I. And it was here, in 1989, that Milosevic held his
infamous rally of 250,000 supporters which launched his campaign
for a Greater Serbia.” 10 With that depth of commitment, it was all
but inconceivable that Milosevic would be talked out of Kosovo by
allied diplomacy, even if supported by a threat of NATO bombing
which he was inclined, for good reason, not to take seriously.11
Expounding further on the erroneousness of assuming that Opera-

tion Allied Force would produce the same relatively quick and easy
results that the earlier Operation Deliberate Force had produced in
the Bosnian crisis of 1995, Adam Roberts noted that “the mythologiz-
ing of [that earlier] campaign ignored one inconvenient fact: that it
followed a period of sharp Serb military reverses on the ground,
______________
10Michael Ignatieff, Virtual War: Kosovo and Beyond, New York, Henry Holt and
Company, Inc., 2000, p. 24.
11 As one observer wrote of Operation Allied Force afterward, “so low was NATO’s
credibility with Milosevic that the threat of war and even war itself were not enough to
convince him that he had anything to fear.” Christopher Cviic, “A Victory All the
Same,” Survival, Summer 2000, p. 178.
including the mass expulsion of the Serbs from the Croatian Krajina.
Also, the 1995 bombing was not against Serbia proper, and thus did
not arouse the same nationalist response as would the bombing in
1999. The real lesson of those 1995 events might be a very different
one: that if NATO wants to have some effect, including through air
power, it needs to have allies among the local belligerents and a
credible land-force component to its strategy.”12 A false assumption
that air power alone had produced the Dayton accords may thus
have contributed further to NATO’s miscalculation that Milosevic
could be induced to give up in Kosovo after merely a few days of to-
ken bombing.13 Aleksa Djilas, son of the Yugoslav cold-war dissident
Milovan Djilas and an able intellectual in his own right, attested from
first-hand knowledge that the West had “badly underestimated the
Serbian attachment to Kosovo.” 14 In light of that, rather than ask
why it took so long for NATO’s bombing to coerce Milosevic to back
down, a more appropriate question might be why he yielded as
quickly as he did.
PROBLEMS AT THE COALITION LEVEL

In their joint statement to the Senate Armed Services Committee af-
ter the war ended, Secretary Cohen and General Shelton rightly
______________
12Adam Roberts, “NATO’s ‘Humanitarian War’ over Kosovo,” Survival, Autumn 1999,
pp. 110–111.
13This point bears emphasizing. It was not just that Serbia’s stakes in Kosovo were
much higher than in Bosnia. The two cases diverged additionally in three fundamen-
tal ways, each of which should logically have led the United States and NATO to adopt
a more robust and considered strategy in the Kosovo war. First, the 1995 NATO air
campaign was linked to a major ground effort by Croatian and Bosnian forces coming
in from the north and west and by some 10,000 NATO troops who had been deployed
weeks prior to the onset of the bombing. In 1999, in contrast, the ground element was
expressly ruled out at the highest levels. Second, the objective of Deliberate Force was
limited (ending the siege of Sarajevo) and achievable through a phased, coercive
bombing campaign, whereas the goals of Allied Force were ambiguous (including
forcing Milosevic back to the bargaining table) and more difficult to achieve through
air power alone. Finally, even before the onset of the 1995 bombing, Milosevic had
told U.S. negotiators that he was interested in forging a deal to end the war in Bosnia
on terms acceptable to the international community. That was anything but the case
on the eve of Allied Force. I thank Ivo Daalder of the Brookings Institution for calling
my attention to these differences.
14Michael Dobbs, “‘Europe’s Last Dictator’ Digs In,” Washington Post, April 26, 1999.
insisted that Operation Allied Force “could not have been conducted
without the NATO alliance and without the infrastructure, transit
and basing access, host-nation force contributions, and most impor-
tant, political and diplomatic support provided by the allies and
other members of the coalition.”15 Yet the conduct of the air war as
an allied effort, however unavoidable it may have been, came at the
cost of a flawed strategy that was further hobbled by the manifold
inefficiencies that were part and parcel of conducting combat opera-
tions by committee.
Those inefficiencies did not take long to manifest themselves. Dur-

ing the air war’s first week, NATO officials reported that up to half of
the proposed strike missions had been aborted due to weather and
“other considerations,” the latter, in many cases, being the refusal of
some allies to approve certain target requests. 16 Indeed, the una-
nimity principle made for a rules-of-engagement regime that often
precluded the efficient use of air power. Beyond that, there was an
understandable lack of U.S. trust in some allies where the most im-
portant sensitivities were concerned. The Pentagon withheld from
the allies mission specifics for literally hundreds of sorties that en-
tailed the use of F-117s, B-2s, and cruise missiles, to ensure strict U.S.
control over those U.S.-only assets and to maintain a firewall against
leaks from any allies who might compromise those operations.17
In addition to the natural friction created by NATO’s committee ap-

proach to target approval, the initial reluctance of its political leaders
to countenance a more aggressive air campaign produced a resound-
ing failure to capitalize on air power’s potential for taking down
entire systems of enemy capability simultaneously. In his first inter-
view after Allied Force had begun six weeks earlier, the air compo-
nent commander, USAF Lieutenant General Michael Short, was
frank in airing his sense of being constrained by the political limits
imposed by NATO, pointing out that the graduated campaign was
______________
15Secretary of Defense William S. Cohen and General Henry H. Shelton, “Joint State-
ment on the Kosovo After-Action Review,” testimony before the Senate Armed Ser-
vices Committee, Washington, D.C., October 14, 1999.
16Thomas W. Lippman and Bradley Graham, “Yugoslavs Fire on U.S. Troops; 3 Miss-
ing,” Washington Post, April 1, 1999.
17Bob Deans, “Pentagon Mum About Air Mission,” European Stars and Stripes, April
27, 1999.
counter to all of his professional instincts.18 Short further admitted

that he was less an architect of the campaign than its implementor.
He was particularly critical of NATO’s unwillingness to threaten a
ground invasion from the start, noting that that failure was making it
doubly difficult for NATO pilots to identify their targets because of
the freedom it had given VJ forces to disperse and hide their tanks
and other vehicles.
Finally, Operation Allied Force was hampered by an inefficient target

planning process. Because NATO had initially expected that the
bombing would last only a few days, it failed to establish a smoothly
running mechanism for target development and review until late
April. The process involved numerous planners in the Pentagon and
elsewhere in the United States, at SHAPE in Belgium, at USEUCOM
headquarters in Stuttgart, Germany, and at the CAOC in Vicenza,
Italy, with each participant logging on daily to the earlier-noted se-
cure digitized military computer network called SIPRNET.
Daily target production began at the U.S. Joint Analysis Center at RAF
Molesworth, England, where analysts collated and transmitted the
latest all-source intelligence, including overhead imagery from satel-
lites and from Air Force Predator, Navy Pioneer, and Army Hunter
UAVs. Because the United States commanded the largest number of
intelligence assets both in the theater and worldwide by a substantial
margin, it proposed most of the targets eventually hit, although other
allies made target nominations as well.19 With the requisite infor-
mation in hand, target planners at SHAPE and USEUCOM would
then begin assembling target folders, conducting assessments of a
proposed target’s military worth, and taking careful looks at the like-
lihood of collateral damage. In addition, lawyers would vet each
proposed target for military significance and for conformity to the
law of armed conflict as reflected in the Geneva Conventions.
Once ready for review and forwarding up the chain of command for
approval, these target nominations would then go to the Joint Target
______________
18 Michael R. Gordon, “Allied Air Chief Stresses Hitting Belgrade Sites,” New York
19General Wesley Clark, USA, testimony to the Senate Armed Services Committee,
Washington, D.C., July 1, 1999.
Coordination Board for final vetting. That board’s recommendations

would then go to Admiral James Ellis, commander of Joint Task Force
Noble Anvil and commander in chief, Allied Force Southern Europe
(CINCSOUTH), and his staff in Naples, who would review all target
nominations and forward his recommendations to General Clark,
who in turn would personally review each target to ensure that it fit
the overall guidelines authorized by the NAC.20
Approved targets would then go back to Admiral Ellis, who would

task both the USAF’s 32nd Air Operations Group at Ramstein Air
Base, Germany, and the 6th Fleet command ship deployed in the
Mediterranean to develop target folders. The 32nd AOG would as-
sign multiple aim points per nominated target set and multiple
weaponeering solutions for a broad spectrum of air-delivered muni-
tions. The 6th Fleet planning staff would do the same for TLAM
targets.21 As one might expect, this exceptionally time-consuming
process greatly limited the number of potential targets that could be
struck at any given time. Moreover, even after these multiple hurdles
had been crossed, an approved target could still be countermanded
or withheld by U.S. or NATO political authorities.22
______________
20John A. Tirpak, “The First Six Weeks,” Air Force Magazine, June 1999, pp. 27–29.
21Dana Priest, “Target Selection Was Long Process,” Washington Post, September 20,
1999.
22Comments on an earlier draft by Hq USAFE/SA, April 6, 2001. As a rule, the 19 indi-
vidual allies did not deliberate over every new target added to the list. True enough,
the NAC—that is, all 19 members, from the United States to Luxembourg—had to
agree to move from one phase in the air war to the next. On January 30, 1999, for ex-
ample, the NAC authorized NATO’s secretary general to commence Phase I (attacking
the IADS and some command and control targets) whenever diplomatic efforts had
been deemed exhausted (as it turned out, on March 24, when Solana finally ordered
Clark to begin the bombing). The NAC also approved moving to Phase II on March 27,
thereby allowing NATO to strike against military targets north of the 44th parallel. Al-
though it never approved Phase III, which entailed strikes against military targets
throughout the former Yugoslavia, the NAC gave de facto approval to entering this
phase on March 30. From that point on, aside from Britain, France, and the United
States, no NATO country ever reviewed, let alone approved or vetoed, any individual
weapon aim point. France insisted on reviewing targets in Montenegro; Britain,
France, and the United States all demanded the right to review any target that had
high political significance or was located in or near civilian areas where the risks of
collateral damage were significant. But the remainder of the allies only got to vote on
proposed new target categories. Moreover, targets struck by U.S. aircraft operating
outside NATO but within USEUCOM were not subject to outside review unless they
met these two criteria.
Further compounding the unavoidable inefficiency of this multistage

and circuitous process, two parallel but separate mechanisms for
mission planning and air tasking were used (see Figure 7.1). As
noted earlier, any U.S.-specific systems involving special sensitivi-
ties, such as the B-2, F-117, and cruise missiles, were allocated by
USEUCOM rather than by NATO, and the CAOC maintained separate
targeting teams for USEUCOM and NATO strike planning. This dual
ATO arrangement meant increased burdens on the planning system
to execute workarounds in cases where automated mission planning
systems could not support the dual process, as well as added compli-
cations in airspace control planning created by the presence of low-
observable aircraft, the limited use of IFF systems in some cases, and
the absence of a single, integrated air picture for all participants. Al-
though the use of stealthy aircraft in this dual-ATO arrangement was
dealt with by time and space deconfliction, it nonetheless made for
problems for allies who were not made privy to those operations, yet
who needed information about them in the interest of their own sit-
uation awareness and force protection.23 Commenting on the fric-
tion that was inevitably occasioned by this cumbersome system,
General Short recalled in hindsight that he was constantly having to
tell allied leaders to “trust me” regarding what U.S. assets would be
doing and that he would have preferred to find a way of ensuring that
the daily allied air operations schedule reflected those U.S. systems
in some usable way. As it was, their absence led on occasion to some
significant force deconfliction problems, such as U.S. aircraft sud-
denly showing up on NATO AWACS displays when and where they
were not expected.24
______________
23This problem will only get worse as the low-observable F-22 and Joint Strike Fighter
begin coming on line in significant numbers toward the end of this decade. Should
the United States intend to use these third-generation stealth aircraft in a coalition
context, as seems to be most likely, a dual ATO arrangement of the type used in Allied
Force will not work. New standardized tactics, techniques, and procedures will need
to be perfected and employed regularly in routine allied and combined peacetime
training. I am grateful to my RAND colleagues James Schneider, Myron Hura, and
Gary McLeod for this important insight.
24John A. Tirpak, “Short’s View of the Air Campaign,” Air Force Magazine, September
1999.
RAND MR1365-7.1
NATO U.S.
Combined Forces Joint Forces
Air Component ATO Process Air Component
Commander’s
Intent
Joint
5th Allied Tactical Air Operations Center
Air Force
Combat Plans
Strategy and
Division
Measures of Merit
Combined Strategic
Air Operations Center Planning
Combat Plans Guidance, Guidance,

Apportionment, Apportionment,
Division
Target Selection Targeting
Master Air Master Air

Attack Planning Attack Planning
ATO Master Air Attack ATO

Production Plan Creation Production
Combat Combat
Operations ATO Production Operations
Division and Distribution Division
Air Tasking Order

Execution
Figure 7.1—Operation Allied Force Planning and Implementation
PROBLEMS AT THE U.S. LEVEL

It was not only the alliance-induced friction that made the air war
inefficient. As Allied Force unfolded, it became increasingly clear
that even the U.S. military component was divided in a high-level
struggle over the most appropriate targeting strategy—a struggle

reminiscent of the feuding that had occurred nine years earlier be-
tween the Army’s corps commanders and the JFACC, USAF Lieu-
tenant General Charles Horner, over the ownership and control of air
operations in Desert Storm.25 There was visible tension in this re-
gard between General Clark and his air commander, General Short,
over the heated issue of target priorities: Aggressive micromanage-
ment on the former’s part was eventually met by understandably
frustrated and increasingly transparent passive-aggressive rebellion
against it on the latter’s. As Clark later characterized this difference
of view in his memoirs, he considered the achievement of success
against Serbian ground troops in the KEZ to be the air effort’s “top
priority,” unlike “some of [his] American commanders [who] sub-
scribed to a more doctrinaire view of the conflict,” one which, he
added, was “the classic view of the American air power adherents
who saw air power as strategically decisive, without recourse to the
dirty business of ground combat,” in contrast to the view of “Army
leaders, who want the Air Force to make a difference on the ground.”
Short, no doubt, would offer his own no-less-principled view of that
characterization.26
Once the initial hope that Milosevic would fold within a few days af-
ter the bombing commenced was proven groundless, NATO was
forced into a scramble to develop an alternative strategy. The im-
mediate result was an internecine battle between Clark and his Air
Force subordinate over where the air attacks should be directed.
Short had naturally chafed from the very beginning at the slowness of
Operation Allied Force to gather momentum—three successive
______________
25For the pertinent details of that controversy, see Benjamin S. Lambeth, The Trans-
formation of American Air Power, Ithaca, New York, Cornell University Press, 2000, pp.
130–138.
26General Wesley K. Clark, Waging Modern War: Bosnia, Kosovo, and the Future of
Combat, New York, Public Affairs, 2001, pp. 241, 243–244. As for charges of “alleged
micromanagement,” Clark said only that he many times “found [himself] working
further down into the details than [he] would have preferred, in an effort to generate
the attack effectiveness against the ground forces that [he] knew we needed.” Ibid., p.
245. Short’s countervailing take on all this is presented in candid detail in Lieutenant
General Michael C. Short, USAF (Ret.), “An Airman’s Lessons from Kosovo,” in John
Andreas Olsen, ed., From Maneuver Warfare to Kosovo, Trondheim, Norway, Royal
Norwegian Air Force Academy, 2001, pp. 257–288.
nights had been required just to get through the 51 targets that had
been approved up to that point, most of them air defense-related
and only a few located anywhere in or near Belgrade.27 In light of the
absence of an allied ground threat to flush out Serbia’s dispersed and
hidden forces in Kosovo, Short insisted that a more effective use of
allied air power would be to pay little heed to those forces and to
concentrate instead on infrastructure targets in and near downtown
Belgrade and other cities, including key electrical power plants and
government ministries.
Indeed, by the account of numerous observers who either partici-

pated in or later watched the videotapes of the 94 top-level video
teleconferences (VTCs) conducted throughout Allied Force, a typical
exchange between Clark and Short during the air war’s early days
would have Clark ask: “Are we bombing those ground forces yet,
Mike?” To which Short would typically offer a noncommital re-
sponse. Even in the case of fixed infrastructure targets, Clark report-
edly would venture deep into the most minute details of the target
list. “Let’s turn to target number 311,” Clark would say, by this ac-
count “opening his binder as other participants flipped to the proper
page, as if they were holding hymnals.” He would then raise ques-
tions about a target’s relevance, expostulate on allied sensitivities, or
abort attacks already in progress. He would also, by this account,
sometimes gainsay his own intelligence experts and targeteers by
looking at a particular DMPI placement and asking “Isn’t that an
apartment building?” or “Can’t we move that [DMPI] over 100 feet?”
At which point Short would be seen “slumping back in his chair,
folding his arms in disgust, and mentally checking out.” General
Jumper would then weigh in out of earshot of the others, and a com-
promise arrangement would typically be worked out. By this in-
formed account, it was never clear to participants whether Clark,
through such ex cathedra interventions, was genuinely responding to
political pressure from above or was engaged in a divide-and-rule
game by playing on putative “constraints” to his advantage and
______________
27Of these initial approved targets, 35 were IADS-related, seven entailed VJ and MUP
facilities, seven involved command and control nodes, and two were industrial.
Comments on an earlier draft by Hq USAFE/HO, May 10, 2001.
gathering diverse inputs and opinions until he heard the one he

wanted to hear.28
As the commander of U.S. naval forces participating in Allied Force,

Vice Admiral Daniel Murphy, recalled after the air war ended,
“there was a fundamental difference of opinion at the outset be-
tween General Clark, who was applying a ground commander’s per-
spective . . . and General Short as to the value of going after fielded
forces.” Short believed that it made little sense to waste valuable
munitions, sorties, and time going after the VJ’s 3rd Army in Kosovo
“if we don’t have an army in the field [or] unless we have defined the
opposing army in the field as a center of gravity.”29 He later com-
mented that he thought going after that elusive army entailed a “high
level-of-effort, high-risk, low-payoff option” because there was no
friendly ground presence poised nearby “to make them pre-
dictable.” 30 Nevertheless, Clark’s view as to where the target priority
emphasis should lie prevailed throughout most of the air war. Not
only did Clark insist on attacking dispersed and hidden VJ ground
forces as the first priority—indisputably his prerogative as the theater
CINC—he reportedly micromanaged the day-to-day execution of Al-
lied Force, at times even choosing the particular type of weapon to
be used against a given target.31
______________
28William M. Arkin, “How Sausage Is Made,” Washington Post, July 17, 2000. Clark
himself later affirmed in a backhanded way that he regarded General Short more as a
subordinate to be managed than as a source of trusted counsel on air employment
matters, and that he looked instead to Short’s immediate Air Force superior, General
Jumper, for the latter: “My real window on the operation was going to be provided by
the senior American airman in Europe, John Jumper. Although he wasn’t in the NATO
chain of command for this operation, as the senior American airman he was my ad-
viser and had all the technology and communications to keep a real-time read on the
operations. As Mike Short’s commander in the American chain of command, he also
had a certain amount of influence in an advisory capacity.” Clark, Waging Modern
War, p. 195.
29Tirpak, “Short’s View of the Air Campaign.”
rope,” February 22, 2000. Short also later indicated his belief that the use of VTCs
“improperly allowed senior leadership to reach down to levels they did not need to be
involved in.”
31In one reported exchange during a daily video teleconference, Clark insisted that
NATO air power remain committed against enemy fielded forces in Kosovo, and Short
countered that such missions were a waste of assets and should be supplanted by
missions against downtown Belgrade. Noting that U.S. aircraft were about to attack
the Serbian special police headquarters in Belgrade, Short said: “This is the jewel in
In fairness to the record, Clark was in the decidedly unenviable posi-

tion of having multiple masters tugging at him from different direc-
tions, including the civilian ambassadors to NATO who made up the
NAC; NATO’s Secretary General Solana, who was responsible for
political control over NATO military operations; and the diverse cast
of players in Washington, notably the president, Secretary Cohen,
General Shelton, and the service chiefs with their independent inter-
ests. In the presence of these often conflicting influences, Clark’s
overarching responsibility as SACEUR was to ensure that coalition
warfare worked and that the allies remained in step until they pro-
duced a successful outcome. To his credit, keeping the other 18 allies
on board to the very end was an immense and remarkable accom-
plishment. As Columbia University political scientist Richard Betts
later pointed out in this respect, Clark’s command “was compro-
mised by more conflicting pressures—political, diplomatic, military,
and legal—than any other in history. Given these constraints, keep-
ing the enterprise from flying apart was no mean feat.”32
That said, Clark had the option all along of leaving the day-to-day
operational responsibilities of planning and implementing the air
effort to his JTF commander, Admiral Ellis, as the principal subordi-
nate warfighting CINC. That is what U.S. Army General George Joul-
wan had done as SACEUR in 1995 with Admiral Leighton Smith dur-
ing Operation Deliberate Force, so he could devote his full time,
attention, and energy to his paramount duties as a diplomat in uni-
form. Instead, Clark elected not only to shoulder his diplomatic bur-
dens as NATO’s supreme commander, but also to conduct the air
war himself from Brussels, in the process bypassing not only Admiral
Ellis but also his air component commander, General Short, in mak-
ing air apportionment decisions. Whereas General H. Norman
Schwarzkopf, on the eve of Desert Storm, had become wholly per-
suaded by his trusted JFACC, then–Lieutenant General Horner, of the
merits of the chosen air campaign strategy, Clark would not be
moved by Short from his less trusting insistence that the VJ’s 3rd
_____________________________________________________________
the crown.” To which Clark replied: “To me, the jewel in the crown is when those B-
52s rumble across Kosovo.” Short: “You and I have known for weeks that we have
different jewelers.” Clark: “My jeweler outranks yours.” Dana Priest, “Tension Grew
with Divide in Strategy,” Washington Post, September 21, 1999.
32Richard K. Betts, “Compromised Command: Inside NATO’s First War,” Foreign Af-
fairs, July/August, 2001, p. 126.
Army in Kosovo, rather than vital equities closer to Milosevic in and

around Belgrade, constituted the principal enemy target set.33
THE DESULTORY ONSET OF THE AIR WAR

Notwithstanding their narrow intent and the admitted constraints
that impeded them, the initial strikes of Allied Force, by their mea-
sured nature, stood in marked contrast to the massed and highly or-
chestrated hammer-blows that were delivered with such paralyzing
effect by coalition air power against Iraq from the earliest moments
of Operation Desert Storm. On the home front, criticism of NATO’s
seeming timidity was both instant and searing. The morning after
the operation’s opening night, Senator John McCain, a former Navy
attack pilot and Vietnam POW, complained that “these bombs are
not going to do the job. . . . It’s almost pathetic. You’re going to so-
lidify the determination of the Serbs to resist a peace agreement.
You’d have to drop the bridges and turn off the lights in Belgrade to
have even a remote chance of changing Milosevic’s mind. What
you’ll get is all the old Vietnam stuff, bombing pauses, escalation,
negotiations, trouble.”34 In a similar vein, NATO’s tentativeness and
preemptive forswearing of a ground option led the respected London
Economist to declare that the West had “stumbled into one of its
riskiest ventures” since World War II and to predict that if the
bombing eventually succeeded, it would “owe as much to luck as to
precision.”35
______________
33During a 10th-anniversary retrospective featuring Schwarzkopf’s principal deputies
in Desert Storm, Horner was emphatic on the crucial importance of the ability of those
key deputies to work together harmoniously in producing the war’s successful
outcome: “The one thing you need to understand if you’re going to understand Desert
Storm is that the relationship among the four people at this table—[Admiral Stanley]
Arthur, [General Walter] Boomer, [Lieutenant General John] Yeosock, and me—was
highly unusual. Such a relationship probably has never existed before, and it probably
won’t exist in the future. The trust and respect we had for one another was
unbelievable. This was a function of personality as much as a desire to get the job
done. Unless you understand our relationships, then you really won’t understand
what went on in Desert Storm, all the good and bad—and there was plenty of each.”
“Ten Years After,” Proceedings, January 2001, p. 65.
34R. W. Apple, Jr., “With Decision to Attack, a New Set of U.S. Goals,” New York Times,
March 25, 1999.
35“Stumbling into War,” The Economist, March 27, 1999, p. 17.
In response to such charges, NATO’s spokesman at the time, RAF Air

Commodore David Wilby, gamely said of the enemy as Allied Force
entered its second week: “He’s hurting. We know that he’s running
short of fuel. We’re starting to hit him very hard on the ground. You
will start to see the resolve starting to crack very quickly.”36 How-
ever, USAF officers were complaining bitterly about the restrictive
rules of engagement from the first days of combat operations. Simi-
larly, RAF pilots flying combat missions out of Italy scored the insipid
air effort as “nancying around” and bordering on cowardice.3 7
General Short later commented that the frustration felt by airmen
was “under control” because the alliance was not losing aircraft and
airmen. He added, however, that had losses begun to occur on a
repetitive basis, the alliance would have had to rethink the guidance
its leaders were handing down on strategy and rules of engage-
ment. 38
Indeed, so counter to military common sense was the strategy se-

lected by NATO that Short became convinced early on that strike
planning was all “just planning for diplomatic threat,” that his air
planners were “just going through the motions to some degree,” and
that “we’re probably never going to drop a bomb.” Short added that
he and his planners had determined that there were somewhere
between 250 and 300 “valid, solid military targets” in the area for the
sort of campaign effort that airmen ideally would like to conduct, but
that he was told: “You’re only going to be allowed to bomb two,
maybe three nights. That’s all Washington can stand, that’s all some
members of the alliance can stand, that’s why you’ve only got 90 tar-
gets, this will all be over in three nights.” At that, Short frankly con-
ceded that he assumed a prior deal had been struck with Milosevic,
whereby Milosevic had told NATO, in effect, that he could not accept
NATO’s terms and keep his job unless NATO bombed him and in-
flicted some degree of at least symbolic damage.39 That meant, or so
Short thought, a token NATO bombing effort against the approved
set of 90 targets, 51 of which were IADS targets selected for force
______________
36James Gerstenzang and Elizabeth Shogren, “Serb TV Airs Footage of 3 Captured U.S.
Soldiers,” Los Angeles Times, April 1, 1999.
37Jonathan Foreman, “The Casualty Myth,” National Review, May 3, 1999, p. 40.
38Short, interview on PBS Frontline.
39Ibid.
protection—both south and north of the 44th parallel—and some in

Montenegro, after which Milosevic would dutifully show the white
flag.
Short later declined even to give Allied Force the courtesy of calling it
a “campaign,” saying that it was not an operation aimed at achieving
clear-cut strategy goals with dispatch, but rather something more in
the nature of “random bombing of military targets.”40 It was one
thing, Short said, to go after enemy tanks and APCs in the Iraqi desert
the way the coalition did with such success in Desert Storm before
the ground offensive began. In that instance, everything behind the
forward edge of the battle area was enemy territory, where one could
attack targets at will without concern for collateral damage or the
potential for killing refugees. In the contrasting case of Kosovo, he
said, “we felt that the risk was enormous, and we felt that we were
going to spend a lot of assets to get minimum return. It was going to
take a lot of sorties to kill a tank, and there was enormous risk of hit-
ting the wrong target because we knew refugees would be moving
around in this ethnic cleansing environment.” Short’s preference
was to “go after the head of the snake,” as he put it. In an illustration
of what he meant, he suggested that ten combat sorties against Bel-
grade would all hit their targets and achieve a desired effect, whereas
“if I send those same ten sorties into Kosovo, perhaps we’ll find a
tank, perhaps not, [and] if we don’t, we send the ten sorties to what
in my business we call a ‘dump target,’ which is a suspected assem-
bly area or a barracks from which the enemy has fled two weeks ago,
and we’ll blow up empty buildings. So the bombs will hit something
but the impact on ethnic cleansing is zero.”
For their part, NATO’s civilian leaders could not even bring them-
selves to face the fact that they were engaged, to all intents and pur-
poses, in an ongoing war. Three weeks into Allied Force, Secretary
Cohen declared before the Senate Armed Services Committee:
“We’re certainly engaged in hostilities. We’re engaged in combat.
Whether that measures up to, quote, a classic definition of war, I’m
not prepared to say.”41 Such diffidence on the administration’s part
was ostensibly intended to reflect due executive-branch obeisance to
______________
40Ibid.
41“Verbatim Special: The Balkan War,” Air Force Magazine, June 1999, p. 50.
the war declaration powers of Congress. Indeed, one report noted

that the White House had expressly ordered all U.S. government
agencies and departments not to refer to ongoing operations as a war
out of concern that by so doing, they might bring the administration
into a confrontation with Congress over war declaration powers.42
Yet the stance also reflected an ingrained administration discomfort
over coming to full grips with what its leaders had signed up for in
Operation Allied Force. That discomfort was most palpably tele-
graphed in President Clinton’s statement on March 26 that the
standoff was “not a conventional thing, where one side’s going to win
and one side’s going to lose.”43
True enough, there was no pronounced groundswell of American

popular support for the Kosovo air war as there had been for the 1991
Gulf War, thanks largely in the latter case to the obvious economic
interests at stake in the Gulf, the blatant cross-border aggression that
characterized Saddam Hussein’s invasion, and President Bush’s
sustained efforts during the preceding five months to mobilize such
support. At the end of the first week, a Washington Post and ABC
News poll found that only 51 percent of the American people ap-
proved of the way President Clinton was handling the Kosovo crisis,
with 55 percent supporting NATO’s air war against Serbia.44 In con-
trast, 79 percent of the American populace had supported the air of-
fensive against Iraq at the start of Operation Desert Storm.45
One can reasonably ask whether NATO’s initial assumptions about

public opinion on the issue of casualties underestimated the degree
of popular support that could have been mobilized for a more robust
and effective strategy by a more proactive and committed U.S.
leadership. The chairman of the respected Louis Harris and Associ-
ates polling firm rejected easy suggestions that the American people
would inevitably oppose the commitment of ground troops or any
other determined use of force. “When the U.S. achieves victory in a
______________
42Bill Gertz and Rowan Scarborough, “No War,” Washington Times, April 16, 1999.
43“Verbatim Special: The Balkan War,” p. 51.
44Charles Babington, “Clinton Sticks with Strikes as Poll Shows 51 Percent in U.S. Ap-
prove,” Washington Post, March 30, 1999.
45Richard Benedetto, “Support Not as High as for Other Strikes,” USA Today, April 2,
1999.
just cause,” he pointed out, “the public applauds the use of force.
When it loses—worse still, when America is defeated or runs away (as
in Somalia or Vietnam)—the public reasonably says the use of the
military was a mistake.” Citing the precedent of Desert Storm, he re-
called how during the days immediately preceding the outbreak of
hostilities, no poll found a majority of Americans in favor of prompt
military action. Yet immediately after the air campaign had begun
and was deemed to have gotten off to a good start, surveys found that
between 68 and 84 percent of those polled approved. Similarly, up to
the day before the Desert Storm ground push commenced, a typical
poll taken by the New York Times and CBS found that the public pre-
ferred a continuation of the air war by 79 percent, with only 11 per-
cent favoring the start of ground operations. A few days after the
ground push began, however, a full 75 percent of those polled be-
lieved it had been “right to start the ground war,” as opposed to only
19 percent who opposed it.46
In contrast to the celebratory reaction and commemorative parades

down Wall Street and Constitution Avenue that predominated in the
heady aftermath of Operation Desert Storm, one reason for the sub-
dued response of the American rank and file to the successful con-
clusion of Allied Force may have been that popular expectations
were so low—limited, at bottom, to the simplest hope that the
United States might somehow extricate itself from the morass it had
entered with its reputation as a superpower still intact. Up to the day
that Milosevic finally caved in, even the most ardent air power pro-
ponents were gloomily eyeing the prospect of an open-ended
bombing campaign. They were also coming to accept the growing
likelihood of having to send in allied ground troops to bring the na-
tion’s involvement to a decisive end. Immediately after the cease-
fire, a USA Today/CNN/Gallup poll reported that 53 percent of
Americans did not consider the outcome to be a victory for the
United States, as opposed to only 40 percent of respondents who did.
The poll further reported 46 percent as believing that worldwide re-
spect for the United States had declined as a result of U.S. actions in
the crisis, as opposed to 44 percent who thought that it had grown.47
______________
46Humphrey Taylor, “Win in Kosovo and the Public Will Approve,” Wall Street Jour-
nal, June 3, 1999.
47James Cox, “Poll: Mission Isn’t Seen as U.S. Victory,” USA Today, June 15, 1999.
THE FAILURE TO EMPLOY A COHERENT PLAN

As noted earlier, everything having to do with arrangements already
in place when Allied Force began was driven by the assumption that
the operation would entail, at most, a two- to three-day series of air
strikes directed at approximately 50 targets. Numerous earlier plan-
ning exercises had generated air attack options that varied in length
from two to roughly ten days. None, however, came close to ap-
proaching anything as protracted as the 78 days that the air effort ul-
timately required. In February 1999, SACEUR directed that all exist-
ing attack plans be interwoven and that two to three days be
assumed as the likely length of expected operations. Taking into ac-
count SACEUR’s guidance (“I’m only going to give you 48 hours”),
the lack of stomach either in the United States or in Europe for a se-
rious combat operation, and the past history of post–Desert Storm
air power application in mere token doses by the Clinton adminis-
tration, virtually no one in the planning loop questioned the short
length of the expected operations.
Once NATO’s hope proved hollow, a frenetic rush ensued at SHAPE

to come up with additional target nominations that could be more
quickly and easily approved by NATO’s political authorities. At the
end of the air war’s first week, Clark had only 100 approved targets.48
With the bombing effort going nowhere, he accordingly went to the
NAC and received blanket approval to go after certain broad classes
of targets, including air defenses, command and control, fielded
forces, and resupply sources, at his own discretion. Other broad tar-
get sets and individual targets of a more politically sensitive nature,
however, still had to be submitted for review by the United States,
Britain, and France.
Having thus been cleared to go after most military targets at will,

Clark pressed his staff to identify 5,000 candidates. His target plan-
ners quickly convinced him that 5,000 legitimate aim points were not
to be found in all of Serbia, whereupon Clark declared a new goal of
coming up with 2,000 target candidates, a goal later derided by some
______________
48Ignatieff, Virtual War, p. 99.
planners as “T2K.”49 That goal soon led to the targeting of objects

that had no connection whatever to Yugoslavia’s military capability,
what William Arkin later characterized as a “mechanical process of
meticulous selection with little true military justification.”50 Some-
times the target selection criterion entailed little more than the fact
that an assigned DMPI was located safely away from civilian homes.
That resulted in an approach to force employment that was “neither
calibrated nor intelligible,” but instead spawned “a succession of un-
focused and unconvincing air excursions—experiments in commu-
nication by detonation.”51 It was only at that point that coalition
planners began a serious and methodical target development pro-
cess, in which prospective targets were categorized into four ascend-
ing tiers of collateral damage sensitivity.
Even then, there was little by way of a consistently applied strategy

behind the target development process. As one U.S. officer reporting
to an assignment at the CAOC midway into the operation noted af-
terward, he was told upon arrival: “I know you won’t believe this, but
we don’t have a plan.” He learned that NATO aircrews could only
attack those targets that came out of the target approval process and
could never, at any time, attack an entire target set systematically in
pursuit of paralysis. Target allocations, he said, were driven by rules
of engagement of the moment, which, in turn, were set primarily on
the basis of judgments regarding what the political traffic would bear
domestically and within the alliance. Whenever an untoward event
occurred that had a negative impact on public opinion, the ROE
would seem to tighten almost reflexively. As a case in point, he
noted, target planners were directed by the “highest levels” to cease
using CBUs after Milosevic’s press staff had persuaded CNN to do a
story on the CBU “terror weapon” that was being employed by
______________
49Elaine M. Grossman, “U.S. Military Debates Link Between Kosovo Air War, Stated
Objectives,” Inside the Pentagon, April 20, 2000, p. 7. According to one Allied Force
participant, Clark would press his operators down the line to propose target candi-
dates. They would reply, “Give us the targets and we will take them out,” to which
Clark countered: “You don’t get it. You develop the targets.” Quoted in Ignatieff, Vir-
tual War, p. 99. Clark himself later justified 2,000 as “a large round number, large
enough to get us past the daily struggle over the number of targets approved for that
day.” Clark, Waging Modern War, p. 250.
50William M. Arkin, “Smart Bombs, Dumb Targeting?” Bulletin of the Atomic Scien-
tists, May/June 2000.
51Ibid.
NATO. 52 In the words of another officer, “nobody ever said, ‘no fool-
ing, what we want to accomplish in this country is X.’” As a result,
NATO started “throwing bombs around, hoping that objectives
would materialize.” Said still another, “the targets we selected—
because we had no objectives—were based on nothing other than
that they had been approved. So we slung lead on targets [but] we
couldn’t say, ‘the objectives are X, so we blew up Y.’”53
Indeed, although the methodology of effects-based targeting had

long since been elevated to a high art, most of the attack planning
throughout Allied Force was not driven by desired effects but rather
entailed simply parceling out sortie and munitions allocations by
target category in boilerplate fashion, without much consideration
given to how neutralizing a target might contribute to advancing the
operation’s objectives. A typical example involved attacking refiner-
ies, factories, and bridges in ones and twos over time rather than as
interconnected components of a larger entity whose simultaneous
destruction might instantly undermine Yugoslavia’s capacity to
function effectively. To be sure, some bridges were dropped not to
curtail the flow of traffic over the bridges, but rather to halt the flow
of commodities that flowed along the river under the bridges, or to
cut fiber-optic cables and other conduits that ran through the
bridges. To that extent, effects-based targeting could be said to have
been successfully applied. For the most part, however, owing to the
absence of any systematic effects-based target analysis and strategy
execution, NATO military chiefs had an unnecessarily hard time
convincing NATO’s civilian leaders of the importance of many tar-
gets. General Jumper scored this failure when he stressed the impor-
tance of effects-based targeting and faulted what often happened
instead, namely, what he called “campaign-by-target-list manage-
ment,” whereby planners simply took a list of approved targets and
managed them on a day-to-day basis. 54
______________
52Personal communication to the author, August 23, 1999.
53Grossman, “U.S. Military Debates Link Between Kosovo Air War, Stated Objectives,”
p. 8.
54Comments at an Air Force Association Eaker Institute colloquy, “Operation Allied
Force: Strategy, Execution, Implications,” held at the Ronald Reagan International
Trade Center, Washington, D.C., August 16, 1999. When asked about effects-based
targeting applications in Allied Force, the former commander of the Joint Warfare
Analysis Center, which provides senior warfighters with the principal analytical
On the plus side, the methodology used in individual target plan-

ning, now a bona fide science in its own right, had evolved to a point
where target analysts could predict, for any given weapon type and
impact angle, how far the blast effects would extend, how far shards
of glass could be expected to fly, and even at what distance they
would retain enough force to penetrate skin. The use of this
methodology in arriving at a precisely determined weapon yield,
aim-point placement, and weapon heading and impact angle to
minimize unwanted collateral damage often proved decisive in per-
suading NATO’s civilian leaders to approve attacks on many of the
most politically sensitive targets. The four-tier collateral damage
predictive model that had been developed toward that end was vali-
dated time and again in strike operations against sensitive targets in
built-up areas. Not only did it permit targeting successes against
electrical power, POL, lines of communication, and other objects of
interest in the very heart of downtown Belgrade, it also allowed for
the planned preservation of systems, such as road links within
Kosovo for later use by KFOR peacekeeping troops.
Nevertheless, the scramble to form a targeting cell and establish

smoother planning procedures in the CAOC spotlighted gross ineffi-
ciencies in the air tasking arrangement. That led General Jumper to
suggest afterward that the Air Force needed to start thinking of the
air operations center “as a weapons system” and giving it the same
seriousness of thought that is now given to weapon systems, recog-
nizing that “our product in war is dead targets, and our product in
peace is all that goes into generating the warrior proficiency that kills
those targets in wartime”—including proficiency at planning and
managing an air campaign.55
After the dust of Operation Allied Force had settled, the since-retired
commandant of the U.S. Marine Corps, General Charles Krulak,
_____________________________________________________________
support for such targeting, remarked, “the campaign was more like random acts of
violence than true effects-based targeting. The legal restrictions and political
constraints in the target approval process were inexplicably given as excuses not to do
effects-based targeting. Achieving the desired effects while minimizing the undesired
effects, particularly under the restrictions and constraints that were placed on
SACEUR, is precisely why effects-based targeting should have been applied. Anything
else is just high-tech vandalism.” Conversation with Captain C. J. Heatley, USN (Ret.),
Arlington, Virginia, June 21, 2000.
55John A. Tirpak, “Kosovo Retrospective,” Air Force Magazine, April 2000, p. 31.
commented from firsthand involvement that “we did not have a real
strategy.”56 Likewise, General Short remarked, in what was surely an
understatement for him, that the bombing effort had produced its
objectives “to some extent by happenstance rather than by de-
sign.”57 There were later intimations that a hidden agenda of both
the Clinton administration and General Clark had been not just a re-
versal of the ethnic cleansing in Kosovo, but nothing less than the
removal of Milosevic from power and the democratization of
Yugoslavia. On that point, one NATO official later described Clark as
having said, “you must understand that the objective is to take
Yugoslavia away from Mr. Milosevic, so we can democratize it and
modernize it. That’s our objective.”58 But it was never communi-
cated to subordinate staffs or made a declared goal of Allied Force.59
Given the unseemly rush for targets that ensued at SHAPE and else-
where for more than a month after NATO’s initial assumptions
proved groundless, it seemed more than a bit disingenuous for ad-
ministration officials to have claimed afterward that although they
had “hoped” that military action would end the Serb abuses in
Kosovo quickly, “we knew that it was equally possible that it would
not and that a sustained campaign might be necessary to stop the
killing and reverse the expulsions” and that “we were prepared to do
what it took to win.”60 In what bore every hallmark of a post-hoc at-
tempt at historical revisionism, one official professed that “people in
Washington” knew that there would be a need to attack infrastruc-
ture targets once it became clear that a three- to four-day bombing
effort would not compel Milosevic to settle, but because the allies,
especially the French, were “not on board” initially, NATO could not
______________
56Grossman, “U.S. Military Debates Link Between Kosovo Air War, Stated Objectives,”
p. 6.
57Tirpak, “Kosovo Retrospective,” p. 33.
58Quoted in Grossman, “U.S. Military Debates Link Between Kosovo Air War, Stated
Objectives,” p. 7.
59General Krulak later remarked that even had it been an unstated goal, it was a “non-
starter,” because it would never have gained the backing of NATO.
60James B. Steinberg, “A Perfect Polemic: Blind to Reality on Kosovo,” Foreign Affairs,
November/December 1999, p. 131.
start attacking Phase III targets until it had consensus about the
bombing.61 Two critics of administration policy countered convinc-
ingly that such claims by the administration that it had been pre-
pared all along for the possible need for a prolonged air campaign
were flatly belied by “the hasty improvisation that marked the
bombing effort.” 62 True enough, General Clark was said on strong
authority never to have suggested that just a few days of bombing
would suffice to do the job, even though he did limit his planners to a
short-duration operations plan out of a conviction that the alliance’s
political leaders would not sit still for anything longer.63 But the pre-
sumptions of both NATO and the most senior officials of the Clinton
administration were well reflected in U.S. interagency reports in Jan-
uary and February 1999, which argued confidently that “after enough
of a defense to sustain his honor and assuage his backers, [Milosevic]
will quickly sue for peace.”64
______________
61This official further claimed that Clinton had never intended to take the ground op-
tion off the table but “downplayed” it at first on the grounds that any public mention
of it could have prompted a bruising debate in Congress and premature pressures to
invoke the War Powers Act. He added that by April, the administration felt compelled
to change that perception when it had become clear that important audiences had
concluded that the president had flatly ruled out any ground option. That attempt to
shift perceptions, he said, included asking Solana to initiate a review of the forces that
would be required and encouraging Clark to accelerate planning for a ground inva-
sion, making no effort to keep this quiet. The official admitted that there was no way a
ground invasion could have been imminent when Milosevic capitulated on June 3, but
that any decision to proceed with an invasion most definitely would have had to be
made by mid-June so that the logistical provisions needed to support a ground offen-
sive could be completed before the onset of winter. Interview by RAND staff, Wash-
ington, D.C., July 11, 2000.
62Christopher Layne and Benjamin Schwarz, “Kosovo II: For the Record,” The Na-
tional Interest, Fall 1999, p. 12.
63Conversation by RAND staff with Lieutenant General Ronald Keys, USAF, director of
operations, U.S. European Command, Stuttgart, Germany, March 8, 2000. Clark him-
self was clear on this point in his subsequently published memoirs. Although he ac-
knowledged that “there was a spirit of hope at the political levels [going into the
bombing] that Milosevic might recognize that NATO was actually going to follow
through with its threat and then quickly concede in order to cut his losses,” he, for his
own part, suspected all along that it was “going to be a long campaign.” Clark, Waging
Modern War, pp. 177, 201.
64Elaine Sciolino and Ethan Bronner, “How a President, Distracted by Scandal, En-
tered Balkan War,” New York Times, April 18, 1999.
THE DOWNSIDE OF ALLIANCE WARFARE

Throughout Operation Allied Force, there were targets that one or
more of the key NATO countries would not approve, those that such
countries would not allow to be hit by attacks launched from their
soil, and those that they would not hit themselves but would allow
other allies to hit. The principal NATO member-states also had dif-
fering political agendas and even differing business and financial in-
terests, which heavily affected their reluctance or unwillingness to
countenance attacks against certain targets. As a result, General
Short was never able to mass forces in the execution of an integrated
campaign plan in pursuit of desired strategic effects that had been
carefully thought through in advance. Instead, he was left to go after
approved targets largely in piecemeal fashion, in what one Allied
Force participant caustically dismissed as “target-based targeting”
rather than conscious effects-based targeting.
As the air war entered its second fitful week, one senior U.S. official
suggested that the bombing effort was turning out to be a real-world
battle laboratory, in which the allies were “learning by doing how
you conduct a NATO operation, both at a political and at a military
level.”65 Another later declared, less charitably: “This is coalition
warfare at its worst.” After Allied Force ended, yet another com-
plained that “the NATO troops had too many political masters. The
system was so cumbersome that it limited the effectiveness of some
of the best technology. Joint STARS, for example, couldn’t be used to
direct aircraft to the targets it saw because it took too long to get ap-
proval for a strike.”66
A senior NATO official commented that “NATO got in way over its
head, stumbled through, didn’t know how to get out, [and] was
______________
65John M. Broder, “How to Lay Doubt Aside and Put the Best Face on a Bad Week in
the Balkans,” New York Times, April 1, 1999.
66David A. Fulghum, “Lessons Learned May Be Flawed,” Aviation Week and Space
Technology, June 14, 1999, p. 205. Even deeper than the problem of slow target ap-
proval, however, was the problem of positive target identification, given the excep-
tional stringency of the prevailing rules of engagement. For example, Joint STARS
could not distinguish a column of refugees from a column of military vehicles loaded
with enemy troops, a performance shortfall far more difficult to fix than streamlining
the approval process. I am grateful to my colleague Bruce Pirnie for pointing this out.
scared to death by what was happening.” This official added that the
entire bombing effort had been a “searing experience” that had “left
a bitter taste of tilting within governments, between governments,
between NATO headquarters in Brussels and the military headquar-
ters at Mons.”67 Reflecting the consensus arrived at by many senior
U.S. military officers, both active and retired, Admiral Leighton
Smith concluded that “the lesson we’ve learned is that coalitions
aren’t good ways to fight wars” and that, at a minimum, the political
process in NATO needed to be streamlined so that the collective
could use force in a way that made greatest military sense.68
In what became a particular sore spot, leaks of target information

were discovered early on during Allied Force, contributing in part to
the change in procedure described above to streamline the target
selection process to allow commanders and planners greater free-
dom to bomb without consulting every NATO ally every time. In one
instance of a suspected leak, two empty Interior Ministry buildings in
Belgrade were struck by cruise missiles at the end of the third week.
Only 24 hours previously, those buildings had been full of employ-
ees, suggesting that the enemy knew the attack was coming and
when.69
Even before that event, the Pentagon had admitted the discovery of
operational security problems, as well as its suspicions that the Serbs
had gained access to at least parts of the ATO, thereby enabling them
to reposition mobile SAMs in anticipation of planned attacks.70 Alle-
gations that France, in particular, had been kept out of the loop with
respect to some target planning because of concern that the infor-
mation would be passed on to Milosevic were tacitly confirmed in
early April by a Clinton confidant, who remarked that “there are cir-
______________
67Jane Perlez, “For Albright’s Mission, More Problems and Risk,” New York Times,
June 7, 1999.
68Bradley Graham and Dana Priest, “‘No Way to Fight a War’: Hard Lessons of Air
Power, Coalitions,” Washington Post, June 6, 1999.
69Hugo Gurdon, “U.S. Admits Milosevic Spies Are Inside NATO,” London Daily Tele-
graph, April 15, 1999.
70 Roberto Suro and Thomas E. Ricks, “Pentagon: Kosovo Air War Data Leaked,”
cles and circles within NATO.” 71 In a post–Allied Force interview,

Clark admitted that at least one ally had leaked secret targeting in-
formation to Yugoslav officials. Without naming the alleged culprit,
he said that the security breach was “as clear as the nose on your
face.”72
After the air war ended, Secretary Cohen conceded in a statement to

the Senate Armed Services Committee that “it was very difficult to
take 19 different countries and get an effective campaign under way
without some bumps in the road.” Cohen added that the alliance
was “slow, in some cases too slow, to achieve a consensus.”73 Citing
what he called “self-inflicted wounds in asymmetric warfare,” Admi-
ral Ellis added, in his own after-action briefing to Pentagon officials,
that the enemy had most definitely drawn aid and comfort from the
cumbersome White House and NAC target approval process, as well
as from the poor operational security the coalition operations had
generated, not only on the NATO side but on the U.S. side as well.74
COMMAND AND CONTROL SHORTCOMINGS

The problems created by the lack of a coherent strategy in Allied
Force were further aggravated by a confusing chain of command, un-
suitable organizational structures, and a lack of staff integration
where it was needed most. Indeed, the air war was dominated by
what General Short called “about as murky a command relationship
as you could possibly get.”75 Two parallel chains of command (see
______________
71 Hugo Gurdon, “France Kept in Dark by Allies,” London Daily Telegraph, April 9,
1999.
72“NATO Chief: Targeting Goals Leaked to Yugoslavia,” Pacific Stars and Stripes, Au-
gust 13, 1999.
73Tom Raum, “Cohen: NATO Process Prolonged Air Strikes,” Philadelphia Inquirer,
July 21, 1999.
74Other examples of such self-inflicted wounds, in Ellis’s view, were excessively high
standards for limiting collateral damage, NATO’s self-suspension of the use of cluster
munitions, the aversion to casualties and ground combat, and the reactive as opposed
to proactive public affairs posture, all of which slowed allied response time and re-
duced allied control over the air war’s operational tempo.
75Quoted in Lieutenant Colonel L. T. Wight, USAF, “What a Tangled Web We Wove:
An After-Action Assessment of Operation Allied Force’s Command and Control Struc-
ture and Processes,” unpublished paper, no date, p. 1. Colonel Wight was a member
of the C-5 Strategy Cell at the CAOC.
Figure 7.2) worked simultaneously for each allied participant: The

first was a NATO chain of command, which began at the North At-
lantic Council, the alliance’s political leadership, and went from
RAND MR1365-7.2
NATO Chain of U.S. Chain of

Command Command
dnaNorth
mmoC Atlantic
lanoitaN National
dnammoCommand
C lanoitaN
seCouncil
itirohtuA Authorities
seitirohtuA
NATO
.S.U U.S.
.S.U
SACEUR General Clark CINCEUR

Supreme Headquarters United States
Allied Powers Europe European Command
Admiral Ellis General Jumper

Allied Forces JTF United States Air
Southern Europe Noble Anvil Forces in Europe
Administrative Control
CFACC Lieutenant General Short JFACC

JTF Noble Anvil
Allied Air Forces
Joint Force
Southern Europe
Air Component
Brigadier General Gelwix 32 AOG and Augmentees

Combined Air Joint Air Operations
Operations Center Center
CAOC and Augmentees

Combat Plans
Division
CAOC and Augmentees

Combat Operations
Division
Figure 7.2—U.S. and Allied Organization for Allied Force

there to General Clark as SACEUR, through the NATO military staff at

SHAPE, to the regionally involved CINCSOUTH, Admiral Ellis, and
his JTF staff in Naples, and finally to General Short as commander,
Allied Air Forces, Southern Europe (COMAIRSOUTH) and his staff,
along with his subordinate Allied Tactical Air Forces, including 5
ATAF, which also operated the CAOC at Vicenza, Italy.
Paralleling this NATO chain of command were the individual chains

of each allied member, typified by that of the United States, which
began with the National Command Authorities (NCA) at the White
House and Pentagon and proceeded to the regional commander in
chief, General Clark, in his capacity as CINCEUR (CINC U.S.
European Command) and, in turn, to the various subordinate U.S.
component commands. The most important two of those subordi-
nate commands were JTF Noble Anvil, established under the com-
mand of Admiral Ellis as CINCSOUTH, and USAFE, under the com-
mand of General Jumper, who retained operational control of some
U.S. assets, specifically the B-1, B-2, B-52, F-117, E-3, KC-135, and U-
2 aircraft that flew in Operation Allied Force. General Short exercised
tactical control over these aircraft and was assigned operational con-
trol of all other combat aircraft assigned to the 31st Air Expeditionary
Wing at Aviano Air Base, Italy.76 Finally, a joint special operations
task force maintained operational control over all aircraft dedicated
to combat search and rescue missions, and the allied participants
ceded operational and tactical control over their aircraft to General
Short, who, in his capacity as COMAIRSOUTH, was the designated
NATO operational commander and who directed all air missions
flown in the NATO-releasable ATO.77
This dual-hatting of so many commanders and operational functions

often made it hard for Allied Force participants, irrespective of level,
to determine exactly who was operating in what capacity at any given
time. For example, the CAOC at Vicenza, which was operated by
______________
76Although the USAF’s still-embryonic Air Expeditionary Forces were not available for
participation in Operation Allied Force, the AEF concept was exercised at Aviano when
the reinforced 31st Fighter Wing was designated a provisional air expeditionary wing
for the air war’s duration.
77Complicating matters even further was the added confusion created by having Task
Force Hawk and JTF Shining Hope functioning as separate command entities within
the joint operating area.
NATO’s 5 ATAF, performed command and control functions both for

NATO and for U.S.-only operations. That odd arrangement em-
anated from the fact that the command and control apparatus put in
place for what ultimately became Operation Allied Force had initially
been created for a U.S.-only operation, an apparatus that remained
in place even as the air war became a NATO effort. As one informed
account of this “flawed organizational structure” later observed, the
JCS, the USAFE staff, and Admiral Ellis’s JTF all “performed roles
outside their doctrinal bounds,” further confusing the execution of
Allied Force and producing numerous instances of “conflicting guid-
ance, command echelons being skipped or omitted entirely, and ei-
ther a duplication of effort or functions not being performed at all,
since one organization erroneously thought the other was responsi-
ble for a particular task.”78
Amplifying further on this bizarre command arrangement, RAF Air

Commodore Andrew Vallance later noted from his vantage point as
chief of the NATO Reaction Forces air staff in Kalkar, Germany, that
the control of an operation in NATO’s southern region would nor-
mally have fallen to CINCSOUTH and his subordinate air comman-
der (COMAIRSOUTH), Admiral Ellis and General Short, as had been
the case earlier with Admiral Smith and then–USAF Lieutenant Gen-
eral Michael Ryan during the successful Operation Deliberate Force
over Bosnia in 1995. Yet in the case of Allied Force, following the
precedent set earlier by NATO’s IFOR/SFOR operation in Bosnia,
General Clark as SACEUR elected to take direct personal control of
the air effort, effectively cutting CINCSOUTH out of the command
chain, to all intents and purposes.
In contrast, the commander, Allied Air Forces, Central Europe

(COMAIRCENT), a USAF four-star general, would not normally have
been directly involved in a southern region operation. However, in
the case of Allied Force, through his national responsibilities as
commander, U.S. Air Forces in Europe (COMUSAFE), General
______________
78Wight, “What a Tangled Web We Wove,” p. 7. As a case in point, the Joint Chiefs,
despite their formal status as advisers to the NCA, issued directives as though they
were part of the warfighting chain of command, for instance, ruling out the use of
CBUs by U.S. forces and placing certain targets on “JCS withhold.” Likewise, JTF No-
ble Anvil was often placed in a position of providing direction and guidance to NATO
operational units, even though it nominally exercised operational and tactical control
over U.S. assets only.
Jumper had “a major say in how the huge USAF contribution was
used.”79 Moreover, unlike General Horner in Desert Storm, who an-
swered directly to the theater CINC, General Schwarzkopf, Short re-
ported not to Clark but rather to Admiral Ellis, who in turn reported
to Clark—a situation which, Short cautiously said, “colors the equa-
tion a bit in terms of my latitude, if you will, in this air campaign.”80
Considering all this confusion and more, concluded an informed and
expert observer, operational effectiveness in Allied Force was prob-
ably achieved “in spite of the . . . command structures and processes
rather than as a direct result of them.”81
In addition, because NATO had initially anticipated that the bomb-

ing would last only a couple of days, the CAOC was woefully under-
staffed and unprepared for the demands that immediately fell upon
it. For example, on the night the air war began, there was no as-
signed strategy cell, no flexible targeting cell, no established guid-
ance, apportionment, and targeting (GAT) process, and no BDA team
in place.
Even when more fully developed, the BDA process left much to be
desired. It was well enough equipped, calling as required on national
and theater offboard sensors such as satellites and the U-2, tactical
sensors such as Predator and Hunter UAVs, and onboard sensors
such as the LANTIRN targeting pod carried by the F-14D, the F-15E,
and the Block 40 F-16CG. Inputs from these information sources
would be forwarded to the JAC at RAF Molesworth, which wielded
chief BDA authority, and other BDA-related entities such as the
CAOC, national agencies, the SACEUR staff, JTF Noble Anvil, and the
JFACC apparatus. However, inputs from national and theater assets
could take days to register an impact because of frequent weather
complications and higher-priority taskings. Moreover, because BDA
______________
79Air Commodore A. G. B. Vallance, RAF, chief of staff, NATO Reaction Forces (Air)
Staff, Kalkar, Germany, “After Kosovo: Implications of Operation Allied Force for Air
Power Development,” unpublished paper, p. 3. Although Clark did, by numerous
eyewitness accounts, sometimes treat Jumper as though he were the air component
commander by virtue of his seniority to Short, Jumper never usurped his superior
rank, never insisted that Short follow his suggestions, and frequently lent a helpful
hand by quietly adjudicating the more prickly VTC sessions to good effect when Clark
and Short got into their differences over targeting strategy and target priorities.
80Michael R. Gordon, “Allied Air Chief Stresses Hitting Belgrade Sites.”
81Wight, “What a Tangled Web We Wove,” p. 1.
often required two or more independent sources to confirm a target

kill, combat assessment often took longer than the time required for
mission planning and retargeting. As a result, targets were often
reattacked unnecessarily, which made for additional operational
inefficiencies, and the air war’s overall progress could not be ade-
quately tracked and measured. For fixed targets in Serbia, BDA con-
firmation was generally adequate, but the results were frequently not
incorporated into replanning. Daily counts of flexible targets known
to have been hit in the KEZ were not kept, resulting in a large band of
uncertainty with respect to estimates of kills of mobile targets. Fi-
nally, there was a recurrent problem with ISR prioritization, reflected
in repeated tension between SACEUR’s tasking of information
sources to support BDA and the felt need at the operator level for
information to support the attacking of targets.82
The CAOC also suffered from an inadequate airspace management

system for assigning tanker tracks and for managing the nightly flow
of combat and combat-support aircraft. Not until late April did the
CAOC create separate flexible targeting cells for enemy IADS assets
and fielded forces. Only by Day 37 was there a smoothly running tar-
get development and review mechanism in place, and only on Day
47 was the first Joint Integrated Prioritized Target List (JIPTL) pro-
duced, along with the first operational assessment briefing to Short
and Clark. Until that time, the would-be Master Air Attack Plan
(MAAP) team had been picking targets almost solely on the basis of
what had been politically approved. That meant that for the first half
of Operation Allied Force, a consistent targeting strategy not only
was not attempted but was not even possible.83
Moreover, the generally poor intelligence preparation of the battle-

field (IPB) occasioned by the faulty assumption that Milosevic would
capitulate after just a few days of token bombing complicated both
planning and execution. NATO’s failure to anticipate and prepare
adequately for a range of adverse enemy actions, such as the com-
mingling of Kosovar Albanian civilians with Serb military convoys
and the highly successful VJ and MUP camouflage, concealment, and
______________
82 Briefing to the author by Brigadier General Daniel J. Darnell, commander, 31st
Fighter Wing, Aviano Air Base, Italy, June 13, 2000.
deception measures, made air operations against both fixed and

mobile targets far more difficult than they had been in Desert Storm.
In addition, IPB in the KEZ was hindered by the absence of a land
component commander in the Allied Force chain of command,
which meant that some of the attendant organizations that could
have helped the JFACC with this mission were also absent.84 On top
of that, the nonstandard target nomination and approval process,
SACEUR’s unusually heavy involvement at the micro-level of target-
ing, and a de facto requirement for zero friendly losses and an abso-
lute minimum of collateral damage hindered the application of clas-
sic doctrinal solutions, limited the choices that were available, and
put extra stress on systems such as UAVs and other ISR assets that
always seemed to be in insufficient supply. Finally, the extended
timelines created by the demands of the target approval process, as
well as the multiplicity of players at senior levels who had managed
to insert themselves into that process, frequently rendered opera-
tions against fleeting targets downright impossible and further at-
tested to the poor integration of ISR management practices with the
command and control functions required to respond within those
timelines. Because the process was so time-consuming, it was fre-
quently impossible to balance the competing priorities of target de-
velopment and battle damage assessment.
Yet another source of friction in the orderly execution of the daily

ATO was the complex overlay of institutional roadblocks and delays,
the net result of which was an information-sharing arrangement
described by one participant as “cumbersome. It really means we
were unable to get timely intelligence to our allies, particularly the
British. . . . It’s not that the information is so secret. It’s that we have
a bureaucracy, and the way we transfer from ‘U.S. Secret’ to ‘NATO
Secret’ takes a little bit of time.”85 As a rule, each allied nation had its
own levels of security classification, and each of these had to be
______________
84The problem was not just the absence of a land component per se, but that no com-
ponent whatsoever undertook the task of IPB until far too late in the operation. What
is required are clearer stipulations regarding whose responsibility it is to conduct IPB,
as well as new approaches and processes for doing so. At present, only the land
component is resourced and prepared to meet that responsibility. Comments on an
earlier draft by Hq USAF/XOXS, July 11, 2001.
85Rowan Scarborough, “Kosovo Target Data Stalled in Transit,” Washington Times,
July 28, 1999.
downgraded in order for the information to be released to other al-

lied participants. Frequently the computer systems that operated
with these different levels were not mutually compatible, and there
were instances, notably in the area of information operations, but
also including B-2 and F-117 operations, in which the very nature of
the activity meant that information could not be widely released.86
Over time, the CAOC went from badly understaffed to packed with a
surfeit of personnel as a result of the rampant inefficiencies of the
target planning and apportionment process. On one occasion, there
were as many as 1,400 people in the small and cramped facility, pro-
ducing a staffing level that bordered on gridlock.87 Some aug-
mentees from other USAF commands brought only limited experi-
ence with high-intensity operations, further hampering the CAOC’s
operational effectiveness. In a representative example of the need-
less inefficiencies that ensued, a PACAF colonel, say, serving as se-
nior duty officer, would overrule something decided at a lower level
with a “we don’t do that in PACAF,” only to have a lieutenant colonel
on the permanent CAOC staff reply, uneasily, that that was the way it
was done in Allied Force, for good reason.88 In general, the abnor-
mally large number of senior officers (lieutenant colonels and
colonels) populating the CAOC limited the effectiveness of the often
more expert junior officers in shaping key decisions. As a rule, the
CAOC and General Short mainly performed battle management and
support functions rather than operating as a master planning center
and high-level command and control entity along the lines of the Air
Operations Center and General Horner in Desert Storm.
Yet for all its eventually ramped-up staffing and improved organiza-
tion, according to its director at the time, the CAOC remained “target
______________
86As for information operations, one Allied Force participant commented that “due to
the involvement of a few compartmented programs, the entire planning effort was
classified at an unnecessarily high level, unreleasable to all but a very few U.S. plan-
ners. Unfortunately, implementing the overall plan was critical to the success of the
operation, but because of the excessive classification, those charged with implement-
ing it could not be told of the plan until it was too late.”
87The CAOC’s normal peacetime manning was around 250 assigned personnel. It had
a reinforced staff of 375 on March 24, the night the air war began, which was finally
ramped up to more than 1,400 as Allied Force peaked at more than 900 sorties a day.
88Conversation with Major General P. J. M. Godderij, deputy commander in chief,
Royal Netherlands Air Force, Scheveningen, the Netherlands, June 7, 2000.
poor” throughout much of Allied Force. Because it was denied any

opportunity to apply an overarching strategy in shaping the air
operation’s plan owing to the slowness and randomness of the target
approval process, “as targets were approved, we’d go hit them. . . .
We had plenty of targets [in principle to go after]—850 or 900—but
no authority to hit them.” Indeed, the CAOC was reportedly so
lacking in available targets and BDA feedback that by Days 55–65,
planners were “putting the same targets up [for approval] two and
three nights in a row, hoping we could give you different DMPIs from
the night before.”89
As for the flexible targeting effort against VJ forces in the KEZ, the
CAOC at first lacked any on-hand Army expertise to help develop the
ground order of battle. With no land component in place, the Army’s
TPQ-36 and TPQ-37 counterbattery radars in Albania required a di-
rect feed to the CAOC, yet information from them was not provided
until the very end because Army doctrine had planned for those sys-
tems to be used in a different manner and the CAOC was not config-
ured to take advantage of them. Worse yet, TF Hawk and its parent
command, the U.S. Army in Europe (USAREUR), consciously elected
not to provide processed intelligence data to the JFACC and JTF
Noble Anvil until circumstances and senior-official intervention
occurred later.90 Eventually, Clark sent a 10-man Army team to the
CAOC to provide such assistance, which aided considerably in the
flexible targeting effort. By mid-May, TF Hawk finally began sending
the CAOC useful real-time targeting information collected by its
counterbattery radars, and a battlefield coordination element staffed
with TF Hawk representatives was established in the CAOC to
provide additional ground intelligence and operator input into the
flexible targeting cell concerned with dispersed and hidden enemy
forces in the KEZ.
______________
90During an after-action presentation by the USAREUR battlefield coordination ele-
ment, Hq USAFE’s AWOS study team learned that JTF Noble Anvil had prepared a
memorandum of agreement for USAREUR coordination expressly stipulating that TF
Hawk would provide the CAOC with processed intelligence data from the TPQ-36 and
TPQ-37 counterbattery radars. In the ensuing coordination process, the USAREUR in-
telligence directorate reportedly excised pertinent language from the text. Comments
on an earlier draft by Hq USAF/XOXS, July 11, 2001.
These ground support elements became progressively more inte-

grated with CAOC operations over time, but their contribution was
disturbingly slow in coming. In his postwar briefing to the Pentagon
leadership, Admiral Ellis suggested that even though no ground op-
eration had been planned for Allied Force, having an assigned joint-
force land component commander in place from the very beginning
would have gone far toward obviating these and most other related
deficiencies.91 There was also a sentiment in the CAOC toward the
end of the air war that the many other units involved in the war ef-
fort, including naval air and the B-2 and F-117 communities, needed
to send their most experienced operators to the CAOC where their
expertise was most badly needed, even if they risked hindering the
operational performance of their parent units as a result. As it was,
the best use of certain systems available to the JFACC was not always
made. For example, the 6th Fleet battle staff consistently felt that its
Carrier Air Wing 8 deployed aboard the USS Theodore Roosevelt was
improperly treated by the CAOC merely as just another allied fighter
squadron, rather than the integrated and independent strike force
with ISR and command-and-control backup it actually was. Navy
planners and operators also pressed repeatedly to have the F-14
TARPS capability employed for direct mission support, whereas the
CAOC persisted in using it primarily for BDA.92
In a widely noted operations management “first,” the use of video

teleconferencing communications was pioneered in Allied Force,
______________
91Amplifying on this point a year after the air war ended, Ellis further remarked that
because air power had been the only force element actively used in Allied Force, the
JFACC naturally had a heavy air emphasis. Yet, he added, the planning and execution
system badly needed land and maritime component commanders deep in the loop as
well, so they could explain to the JFACC, as authoritative equals, what their services
were able to bring to the planning table. Noting how the “J” in JFACC was all too often
silent, Ellis recalled that the contributions of other services were not invariably made
the best use of. For example, he said, the EA-6B, TLAM, and F-14 TARPS all brought
good capabilities to the fight and the JFACC needed to know about those capabilities
directly from their most senior operators. TARPS, in particular, offered excellent po-
tential value, but the Air Force, now out of the manned tactical reconnaissance busi-
ness, sometimes gave the impression of believing that if the information did not come
from space, it did not have an obvious use. Ellis’s overall point was that the services
have not yet become sufficiently joint-minded at the operational and tactical levels, let
alone the strategic level. Interview with Admiral Ellis, May 30, 2000.
92 Conversation with Vice Admiral Daniel J. Murphy, USN, 6th Fleet commander,
aboard the USS LaSalle, Gaeta, Italy, June 8, 2000.
with VTC sessions taking place daily at the most senior level because
of the wide geographic spread of the key players. Sometimes as
many as three or four VTCs were conducted in one day among the
most senior principals. Admiral Ellis later characterized them as a
powerful tool if properly used, owing to their ability to shorten deci-
sion cycle times dramatically, to communicate a commander’s intent
clearly and unambiguously, and to obviate any requirement for the
leading commanders to be collocated. But he cited the propensity of
VTCs to be voracious consumers of leadership and staff working
hours (often involving time wasted composing flashy but unneces-
sary—and even at times counterproductive—briefing graphics) and
poor substitutes for rigorous mission planning and written orders.
Decisions made in the VTC were all too readily prone to misinterpre-
tation as key guidance was successively handed down to lower staff
levels.93
Indeed, in contrast to Desert Storm, the ad hoc nature of the initial

planning, the absence of collocation of senior commanders, the
highly distributed nature of the bombing effort, the compartmented
and often overclassified planning, and an overreliance on email,
VTCs, and other undocumented communication resulted in a no-
table lack of integration of many of the key staff elements in Allied
Force. Typically the only time General Clark was able to speak to his
subordinate commanders was via the daily VTC, a limitation that one
observer said “made it extremely difficult for the senior leaders to
develop a useful working relationship where they possessed the nec-
essary trust and confidence to issue and execute ‘mission-type’ or-
ders without the need to provide detailed tactical guidance.”9 4
Clark’s VTC guidance was never written down or distributed in any
systematic way. In the absence of such formal documentation, most
cell chiefs did their best to debrief their staffs. Yet the time-pressures
of combat frequently made doing that nigh impossible, with the re-
sult that “rumor guidance” tended to predominate throughout the
course of Allied Force.95
______________
93Briefing by Admiral James O. Ellis, USN, commander in chief, U.S. Naval Forces,
Anvil, “The View from the Top,” 1999.
95Ibid., p. 11.
After the war ended, criticism of the VTC approach by many senior
officers was quite vocal. In a characteristic observation, the UK
Ministry of Defense’s director of operations in Allied Force, RAF Air
Marshal Sir John Day, remarked that for all its admitted efficiencies
when its use was properly disciplined, the VTC mechanism was
highly conducive to “ad-hocracy” of all sorts, sometimes resulting in
a lack of clarity regarding important matters of both planning and
execution. For example, he observed that because of the federated
nature of the operation’s planning and the extensive use of VTCs in-
volving a large number of U.S. and NATO headquarters, many agen-
cies had full knowledge of the planning details. That generated initial
confusion among the UK participants as to who precisely was
running the air war, since, until it was confirmed (as suspected) that
it was indeed General Short, they could obtain the same information
from any headquarters that was involved in the VTC. Consistent with
others who reflected on the many negatives of VTCs with the benefit
of hindsight, Air Marshal Day suggested that participation in high-
level VTCs should henceforth be limited exclusively to those directly
in the chain of command and that the commander in chief should
devote careful thought beforehand to the following: (1) the appro-
priate participants and viewers; (2) a prior agenda, so that essential
participants would not hesitate to raise an item out of fear that an
item might already be on the CINC’s checklist; (3) diligent minute-
taking; and (4) a summary of command decisions taken, so that the
commander’s intent would always be unambiguous.96
______________
96Interview with Air Marshal Sir John Day, RAF Innsworth, United Kingdom, July 26,
2000. Rather more bluntly, retired USAF General Chuck Horner, the JFACC during
Desert Storm, commented that had he been SACEUR during Allied Force, he would
have shot every TV monitor in sight. The biggest problem with VTCs, Horner said, is
that one does not know who is present and listening, even as a videotaped record of
the proceeding is being made. That, he added, inclines participants to pull their
punches and speak “for the record,” rather than to speak their mind in a manner that
only privacy can ensure. Conversation with General Horner at Farnborough, United
Kingdom, July 27, 2000.
Chapter Eight
NATO’S AIR WAR IN PERSPECTIVE
Operation Allied Force was the most intense and sustained military
operation to have been conducted in Europe since the end of World
War II. It represented the first extended use of military force by
NATO, as well as the first major combat operation conducted for
humanitarian objectives against a state committing atrocities within
its own borders. It was the longest U.S. combat operation to have
taken place since the war in Vietnam, which ended in 1975. At a
price tag of more than $3 billion all told, it was also a notably expen-
sive one.1 Yet in part precisely because of that investment, it turned
out to have been an unprecedented exercise in the discriminate use
of force on a large scale. Although there were some unfortunate and
highly publicized cases in which innocent civilians were tragically
killed, Secretary of Defense William Cohen was on point when he
characterized Allied Force afterward as “the most precise application
of air power in history.” 2 In all, out of some 28,000 high-explosive
munitions expended altogether over the air war’s 78-day course, no
more than 500 noncombatants in Serbia and Kosovo died as a direct
result of errant air attacks, a new low in American wartime experi-
ence when compared to both Vietnam and Desert Storm.3
______________
1Lisa Hoffman, “U.S. Taxpayers Faced with Mounting Kosovo War Costs,” Washington
2Bradley Graham, “Air Power ‘Effective, Successful,’ Cohen Says,” Washington Post,
June 11, 1999.
3That was the final assessment of an unofficial post–Allied Force bomb damage survey
conducted in Serbia, Kosovo, and Montenegro by a team of inspectors representing
Human Rights Watch. A U.S. Air Force analyst who was later briefed on the study
commented that Human Rights Watch had “the best on-the-ground data of anyone in
219
After Allied Force ended, air power’s detractors lost no time in seek-
ing to deprecate NATO’s achievement. In a representative case in
point, retired U.S. Army Lieutenant General William Odom charged
that “this war didn’t do anything to vindicate air power. It didn’t
stop the ethnic cleansing, and it didn’t remove Milosevic”—as
though those were ever the expected goals of NATO’s air power em-
ployment to begin with. 4 Yet because of the air war’s ultimate suc-
cess in forcing Milosevic to yield to NATO’s demands, the predomi-
nant tendency among most outside observers was to characterize it
as a watershed achievement for air power. One account called Op-
eration Allied Force “one of history’s most impressive air cam-
paigns.”5 Another suggested that if the cease-fire held, the United
States and its allies would have accomplished “what some military
experts had predicted was impossible: a victory achieved with air
power alone.”6 A Wall Street Journal article declared that Milosevic’s
capitulation had marked “one of the biggest victories ever for air
power,” finally vindicating the long-proclaimed belief of airmen that
“air power alone can win some kind of victory.”7 And the New York
_____________________________________________________________
the West.” “A New Bomb Damage Report,” Newsweek, December 20, 1999, p. 4. A
later report, however, indicated that Human Rights Watch had identified 90 separate
collateral damage incidents, in contrast to the acknowledgment by NATO and the U.S.
government of only 20 to 30. Bradley Graham, “Report Says NATO Bombing Killed 500
Civilians in Yugoslavia,” Washington Post, February 7, 2000.
4Mark Thompson, “Warfighting 101,” Time, June 14, 1999, p. 50. Regarding Odom’s
first charge, General Jumper categorically declared after the bombing effort success-
fully ended that “no airman ever promised that air power would stop the genocide
that was already ongoing by the time we were allowed to start this campaign.” Quoted
in The Air War Over Serbia: Aerospace Power in Operation Allied Force, Washington,
D.C., Hq United States Air Force, April 1, 2000, p. 19. One of the few detractors of air
power who was later moved to offer an apologia for having been wrong was military
historian John Keegan, who acknowledged a week before Milosevic finally capitulated
that he felt “rather as a creationist Christian . . . being shown his first dinosaur bone.”
John Keegan, “Modern Weapons Hit War Wisdom,” Sydney Morning Herald, June 5,
1999. Keegan, long a skeptic of air power’s avowed promise, wrote on the eve of
Milosevic’s capitulation that the looming settlement represented “a victory for air
power and air power alone.” Quoted in Elliott Abrams, “Just War. Just Means?” Na-
tional Review, June 28, 1999, p. 16.
5William Drozdiak and Anne Swardson, “Military, Diplomatic Offensives Bring About
Accord,” Washington Post, June 4, 1999.
6Paul Richter, “Air-Only Campaign Offers a False Sense of Security, Some Say,” Los
Angeles Times, June 4, 1999.
7Thomas E. Ricks and Anne Marie Squeo, “Kosovo Campaign Showcased the Effec-
tiveness of Air Power,” Wall Street Journal, June 4, 1999.
NATO’s Air War in Perspective 221
Times called the operation’s outcome “a success and more—a refu-

tation of the common wisdom that air power alone could never make
a despot back down.”8 These and similar views were aired by many
of the same American newspapers that, for the preceding 11 weeks,
had doubted whether NATO’s strategy would ever succeed without
an accompanying ground invasion.
Similarly, defense analyst Andrew Krepinevich, a frequent critic of

claims made by air power proponents, conceded that “almost alone,
American air power broke the back of the Yugoslav military and
forced Slobodan Milosevic to yield to NATO’s demands. What air
power accomplished in Operation Allied Force would have been in-
conceivable to most military experts 15 years ago.” Krepinevich fur-
ther acknowledged that unlike earlier times when air power was
considered by other services to be merely a support element for land
and maritime operations, that was no longer the case today, since air
power had clearly demonstrated its ability in Allied Force to “move
beyond the supporting role to become an equal (and sometimes
dominant) partner with the land and maritime forces.”9
It was not just outside observers, moreover, who gave such ready
voice to that upbeat assessment. Shortly after the cease-fire, Presi-
dent Clinton himself declared that the outcome of Allied Force
“proved that a sustained air campaign, under the right conditions,
can stop an army on the ground.”10 Other administration leaders
were equally quick to congratulate air power for what it had done to
salvage a situation that looked, almost until the last moment, as
though it was headed nowhere but to a NATO ground involvement of
some sort. In their joint statement to the Senate Armed Services
Committee after the air war ended, Secretary Cohen and General
______________
8Serge Schmemann, “Now, Onward to the Next Kosovo. If There Is One,” New York
9Andrew Krepinevich, “Two Cheers for Air Power,” Wall Street Journal, June 11, 1999.
10Pat Towell, “Lawmakers Urge Armed Forces to Focus on High-Tech Future,” Con-
gressional Quarterly Weekly, June 26, 1999, p. 1564. Actually, the air effort proved no
such thing with respect to VJ forces operating in Kosovo.
Henry Shelton, the chairman of the JCS, described it as “an over-

whelming success.”11
With all due respect for the unmatched professionalism of those al-
lied aircrews who, against difficult odds, actually carried out the air
effort and made it succeed in the end, it is hard to accept such glow-
ing characterizations as the proper conclusions to be drawn from
Allied Force. In fact, many of them are at marked odds with the
views of those senior professionals who, one would think, would be
most familiar with air power and its limitations. Shortly before the
bombing effort began, the four U.S. service chiefs uniformly
doubted, in testimony before the Senate Armed Services Committee,
whether air strikes by themselves would succeed in compelling Milo-
sevic to yield.12 Indeed, the Air Force chief of staff, General Michael
Ryan, admitted less than a week later: “I don’t know if we can do it
without ground troops.”13 After Allied Force was over, the former
commander of NATO forces during Operation Deliberate Force, Ad-
miral Leighton Smith, remarked that the Kosovo experience should
go down as “possibly the worst way we employed our military forces
in history.” Smith added that telling the enemy beforehand what you
are not going to do is “the absolutely dumbest thing you can do.” 14
Former Air Force chief of staff General Ronald Fogleman likewise ob-
served that “just because it comes out reasonably well, at least in the
eyes of the administration, doesn’t mean it was conducted properly.
The application of air power was flawed.” Finally, the air component
commander, USAF Lieutenant General Michael Short, declared that
“as an airman, I’d have done this a whole lot differently than I was
allowed to do. We could have done this differently. We should have
done this differently.”15
______________
11Secretary of Defense William S. Cohen and General Henry H. Shelton, “Joint State-
ment on the Kosovo After-Action Review,” testimony before the Senate Armed
Services Committee, Washington, D.C., October 14, 1999.
12 Bradley Graham, “Joint Chiefs Doubted Air Strategy,” Washington Post, April 5,
1999.
13Quoted in “Verbatim Special: The Balkan War,” Air Force Magazine, June 1999, p.
47.
14“Reporters’ Notebook,” Defense Week, July 19, 1999, p. 4.
15William Drozdiak, “Allies Need Upgrade, General Says,” Washington Post, June 20,
1999.
Indeed, few Allied Force participants were more surprised by the

sudden capitulation of Milosevic than the majority of the alliance’s
most senior airmen.16 By the end of May, most USAF generals had
concluded that NATO would be unable to find and destroy any more
dispersed VJ troops and equipment without incurring more unin-
tended civilian casualties.17 General Short had reluctantly con-
cluded that NATO’s strategy, at its existing level of intensity, was
unlikely to break Milosevic’s will and that there was a clear need to
ramp up the bombing effort if the alliance was to prevail. 18 True
enough, on the eve of the cease-fire, General Ryan predicted that
once the air effort began seeking strategic rather than merely battle-
field effects, Milosevic would wake up to the realization that NATO
was taking his country apart on the installment plan and that his ul-
timate defeat was “inevitable.” The Air Force chief hastened to add,
however, that Allied Force had not begun in “the way that America
normally would apply air power,” implying his belief that there was a
more sensible way of going about it.19 As a testament to widespread
doubts that the air war was anywhere close to achieving its objec-
tives, planning was under way for a continuation of offensive air op-
erations against Yugoslavia through December or longer if neces-
sary—although it remains doubtful whether popular support on ei-
ther side of the Atlantic would have sustained operations for that
long.
In sum, Operation Allied Force was a mixed experience for the

United States and NATO. Although it represented a successful appli-
cation of air power in the end, it also was a less-than-exemplary ex-
______________
16Most others as well were caught off guard by the sudden ending of the Kosovo crisis.
See Lieutenant General Bernard E. Trainor, USMC (Ret.), “The Council on Foreign
Relations Report on the Kosovo Air Campaign: A Digest of the Roundtable on the Air
Campaign in the Balkans,” Council on Foreign Relations, New York, July 27, 2000. One
notable exception was USAF Brigadier General Daniel J. Leaf, commander of the 31st
Air Expeditionary Wing at Aviano Air Base, Italy, who confidently told his aircrews on
the eve of Milosevic’s capitulation that he could “smell an impending NATO victory in
the air” (conversation with the author in Washington, D.C., November 16, 2000).
17John F. Harris and Bradley Graham, “Clinton Is Reassessing Sufficiency of Air War,”
18William M. Arkin, “Limited Warfare in Kosovo Not Working,” Seattle Times, May 22,
1999.
19General Michael E. Ryan, “Air Power Is Working in Kosovo,” Washington Post, June
4, 1999.
ercise in strategy and an object lesson in the limitations of alliance

warfare. Accordingly, any balanced appraisal of the operation must
account not only for its signal accomplishments, but also for its
shortcomings in both planning and execution, which came close to
making it a disaster for the alliance.
THE ACHIEVEMENTS OF ALLIED FORCE

Admittedly, there is much to be said of a positive nature about
NATO’s air war for Kosovo. To begin with, it did indeed represent
the first time in which air power coerced an enemy leader to yield
with no friendly land combat action whatsoever.20 In that respect,
the air effort’s conduct and results well bore out a subsequent obser-
vation by Australian air power historian Alan Stephens that “modern
war is concerned more with acceptable political outcomes than with
seizing and holding ground.”21
It hardly follows from this, of course, that air power can now “win
wars alone” or that the air-only strategy ultimately adopted by the
Clinton administration and NATO’s political leaders was the wisest
choice available to them. Yet the fact that air power prevailed on its
own despite the multiple drawbacks of a reluctant administration, a
divided Congress, an indifferent public, a potentially fractious al-
liance, a determined enemy, and, not least, the absence of a credible
NATO strategy surely testified that the air weapon has come a long
way in recent years in its relative combat leverage compared to other,
more traditional force elements. Thanks to the marked improve-
ments in precision attack and battlespace awareness, unintended
damage to civilian structures and noncombatant fatalities were kept
to a minimum, even as air power plainly demonstrated its coercive
potential.
______________
20It bears noting here that the December 1972 bombing of Hanoi was also an example
of successful coercive bombing, albeit with a very limited objective and in the context
of a much larger war that ended in defeat for the United States. For more on this, see
Wayne Thompson, To Hanoi and Back: The U.S. Air Force and North Vietnam, 1966–
1973, Washington, D.C., Smithsonian Institute Press, 2000, pp. 255–280.
21Alan Stephens, Kosovo, or the Future of War, Paper Number 77, Air Power Studies
Center, Royal Australian Air Force, Fairbairn, Australia, August 1999, p. 21.
In contrast to Desert Storm, the air war’s attempts at denial did not
bear much fruit in the end. Allied air attacks against dispersed and
hidden enemy forces were largely ineffective, in considerable part
because of the decision made by NATO’s leaders at the outset to
forgo even the threat of a ground invasion. Hence, Serb atrocities
against the Kosovar Albanians increased even as NATO air opera-
tions intensified. Yet ironically, in contrast to the coalition’s ulti-
mately unsuccessful efforts to coerce Saddam Hussein into submis-
sion, punishment did seem to work against Milosevic, disconfirming
the common adage that air power can beat up on an adversary indef-
initely but rarely can induce him to change his mind.
Although these and other operational and tactical achievements

were notable in and of themselves and offered ample grist for the
Kosovo “lessons learned” mill, the most important accomplishments
of Allied Force occurred at the strategic level and had to do with the
performance of the alliance as a combat collective. First, notwith-
standing the charges of some critics to the contrary, NATO clearly
prevailed over Milosevic in the end. In the early aftermath of the air
war, more than a few observers hastened to suggest that NATO’s
bombing had actually caused precisely what it had sought to prevent.
Political scientist Michael Mandelbaum, for example, portrayed Al-
lied Force as “a military success and political failure,” charging that
while it admittedly forced a Serb withdrawal from Kosovo, the
broader consequences were the opposite of what NATO’s chiefs had
intended because the Kosovar Albanians “emerged from the war
considerably worse off than they had been before.”22 Another charge
voiced by some was that as Allied Force wore on, NATO watered
down the demands it had initially levied on Milosevic at Rambouillet.
As early as the air war’s 12th day, this charge noted, NATO merely
stipulated that Kosovo must be under the protection of an
______________
22Michael Mandelbaum, “A Perfect Failure: NATO’s War Against Yugoslavia,” Foreign
Affairs, September/October 1999, p. 2. That charge was based on the fact that prior to
the air war’s start on March 24, 1999, only some 2,500 civilian innocents had died in
the Serb-Albanian civil war, whereas during the 11-week bombing effort, an estimated
10,000 civilians were killed by marauding bands of Serbs unleashed by Milosevic in di-
rect response to Allied Force.
“international” security force, whereas at Rambouillet, it had insisted

on that presence being a NATO force.23
There is no denying that the Serb ethnic cleansing push accelerated

after Operation Allied Force began. It is even likely that the air effort
was a major, if not determining, factor behind that acceleration. Yet
it seems equally likely that some form of Operation Horseshoe, as the
ethnic cleansing campaign was code-named, would have been un-
leashed by Milosevic in any event during the spring or summer of
1999. Indeed, what a Serb general was later said by SACEUR to have
forecast as a “hot spring” in which “the problem of Kosovo . . . will
definitely be solved” commenced more that a week before the start of
Allied Force, when VJ and MUP strength in and around Kosovo was
increased by 42,000 troops and some 1,000 heavy weapons—even as
the Rambouillet talks were under way.24 Administration defenders
are on solid ground in insisting that the ethnic cleansing had already
begun and that had NATO not finally acted when it did, upward of
a million Kosovar refugees may well have been left stranded in
Albania, Macedonia, and Montenegro, with no hope of returning
home.25
Although NATO’s air strikes were unable to halt Milosevic’s ethnic

cleansing campaign before it had been essentially completed, they
did succeed in completely reversing its effects in the early aftermath
of the cease-fire. Within two weeks of the air war’s conclusion, more
than 600,000 of the nearly 800,000 ethnic Albanian and other
refugees had returned home. By the end of July, barely one month
after the cease-fire, only some 50,000 displaced Kosovar Albanians
still awaited repatriation (see Figure 8.1). By any reasonable mea-
sure, Milosevic’s bowing to NATO reflected a defeat on his part, and
his accession to the cease-fire left him worse off than he would have
been had he accepted NATO’s conditions at Rambouillet. Under the
______________
23Robert Hewson, “Operation Allied Force: The First 30 Days,” World Air Power Jour-
nal, Fall 1999, p. 24.
24“Briefing by SACEUR General Wesley Clark,” Brussels, NATO Headquarters, April
13, 1999.
25 See, for example, the riposte to Mandelbaum by the Clinton administration’s
deputy national security adviser, James B. Steinberg, “A Perfect Polemic: Blind to
Reality on Kosovo,” Foreign Affairs, November/December 1999, pp. 128–133.
RAND MR1365-8.1
800
700
Total
600
500
Thousands
Albania
400
300
Macedonia
200
100 Bosnia Montenegro
0
23
20
18
15
29
13
27
1
ril
ay
ne
st
ch
ril
ay
ne
ly
ly
gu
Ap
Ju
Ju
Ju
Ap
M
ar
Ju
Ju
Au
M
SOURCE: UN High Commissioner for Refugees, “Kosovo Crisis Update,”

March 31, 1999.
Figure 8.1—Refugee Flow
terms of Rambouillet, Serbia would have been permitted to keep

5,000 of its “security forces” in Kosovo. Thanks to the settlement ul-
timately reached before the cease-fire, however, there are now none.
Moreover, on the eve of Operation Allied Force, Milosevic had in-
sisted as a point of principle that not a single foreign troop would be
allowed to set foot on Kosovo soil. Today, with some 42,000 KFOR
soldiers from 39 countries performing daily peacekeeping functions,
Kosovo is an international protectorate safeguarded both by the UN
and NATO, rendering any continued Serb claim to sovereignty over
the province a polite fiction. At bottom, as NATO’s Secretary Gen-
eral, Javier Solana, declared in a retrospective commentary on the
experience, the alliance “achieved every one of its goals” in forcing a
Serb withdrawal from Kosovo.26 Whether or not one chooses to call

that outcome a “victory” entails what Karl Mueller has characterized
as “a semantic exercise that should only really matter to social scien-
tists seeking to code the event for data analysis.”27
Second, NATO showed that it could operate successfully under pres-

sure as an alliance, even in the face of constant hesitancy and reluc-
tance on the part of many of the member-states’ political leaders.
For all the air war’s fits and starts and the manifold frustrations they
caused, the alliance earned justified credit for having done remark-
ably well in a uniquely challenging situation. In seeing Allied Force
to a successful conclusion, NATO did something that it had been
neither created nor configured to do. Indeed, it might well have been
easier for Washington and SACEUR to elicit NAC approval to grant
border-crossing authority at the brink of a NATO–Warsaw Pact
showdown during the height of the cold war than to get 19 post–cold
war players on board for an offensive operation conducted to ad-
dress a problem that threatened no member’s most vital security in-
terests. As General Clark later recalled, the “ultimate proof” of the air
war’s success was that NATO realized its “ability to maintain alliance
cohesion despite all the pressures of fighting a conflict, at the same
time bringing in new members, and then going into Kosovo itself on
an extended and uncertain campaign—uncertain in that there [was]
no fixed exit date.”28
Reflecting on the air war experience a year later, Admiral James Ellis,
the commander of the U.S. contribution to Allied Force, observed
that during the final days leading up to March 24, it was a question
not of how the bombing effort would be conducted so much as
whether it would take place at all. Before Rambouillet, the challenge
had been to compel Milosevic to do something. Afterward, it became
to compel him to stop doing something. Ellis speculated that had the
allies known from the outset that they were signing up for a 78-day
campaign, they might easily have declined the opportunity forth-
______________
26Javier Solana, “NATO’s Success in Kosovo,” Foreign Affairs, November/December
1999, p. 114.
27Karl Mueller, “Deus ex Machina? Coercive Air Power in Bosnia and Kosovo,” un-
published paper, School of Advanced Air Power Studies, Maxwell AFB, Alabama,
November 7, 1999, p. 6.
28“Wesley Clark Looks Back,” National Journal, February 26, 2000, p. 612.
with. Unlike the ad hoc group of nations that fought Desert Storm as
a solidly united front, NATO was not a coalition of the willing but
rather a loose defensive alliance of 19 democracies. They were all
strongly inclined to march to different drummers, and all had varying
commitments to grappling—at least militarily—with humanitarian
crises in which they had no clear national security stake.29
As the bombing entered its third month without a clear end in sight,
Ellis feared that allied cohesion might collapse within three weeks
unless something of a game-changing nature occurred, such as a
drastic move by Milosevic to alter the stakes or a firm U.S. decision to
accede to a ground-invasion option. Offsetting that fear, however,
was his belief that the allies were finally beginning to recognize and
accept the need to come to terms with some thorny operational is-
sues such as granting approval to attack electrical power and other
key infrastructure targets. That took time, Ellis said, but the fact that
it finally occurred constituted a signal that the alliance was slowly
learning how to do what needed to be done.
Finally, for all the criticism that was directed against some of the less
steadfast NATO members for their rear-guard resistance and ques-
tionable loyalty while the air war was under way, even the Greek gov-
ernment held firm to the very end, despite the fact that more than 90
percent of the Greek population supported the Serbs rather than the
Kosovar Albanians—and held frequent large-scale street demonstra-
tions to show that support. 30 True enough, there remain many un-
knowns about the outlook for NATO’s steadfastness in any future
confrontation along Europe’s eastern periphery. Yet NATO was able
to maintain the one quality that was essential for the success of Allied
______________
29 Interview with Admiral James O. Ellis, USN, commander in chief, Allied Forces,
Southern Europe, Naples, Italy, May 30, 2000. This is not to say, however, that the al-
lies had no intrinsic stake at all. Italy had a stake in preventing further depredations
by Milosevic because of the refugee problem they created. Greece had a major stake
in what happened to the Serbs because of a largely sympathetic population. Germany
also found itself being inundated with refugees. Hungary had good reason to worry
about the Hungarian population still inside Serbia. All of the NATO countries had an
intrinsic interest in stability in Europe, and Milosevic was, if nothing else, a destabi-
lizer of the first order. I am grateful to Alan Gropman of the Industrial College of the
Armed Forces, Washington, D.C., for reminding me of these important facts.
30Air Commodore A. G. B. Vallance, RAF, chief of staff, NATO Reaction Forces (Air)
Staff, Kalkar, Germany, “Did We Really Have a Good War? Myths in the Making,” un-
published manuscript, no date, p. 2.
Force: its cohesion and integrity as a fighting collective. The lion’s

share of the credit for that, suggested Air Marshal Sir John Day, be-
longs to NATO Secretary General Solana, who, in what Day called a
“brilliant” performance, showed both leadership and courage in the
face of continuous U.S. pushing and an equally continuous reluc-
tance on the part of many allies to go along.31
THE AIR WAR’S FAILINGS

Despite these accomplishments, enough discomfiting surprises em-
anated from the Allied Force experience to suggest that instead of
basking in the glow of air power’s largely single-handed successful
performance, air warfare professionals should give careful thought to
the hard work that still needs to be done to realize air power’s fullest
potential in joint warfare. As in the case of the various positive out-
comes noted above, many of these surprises entailed shortfalls at the
tactical and operational levels. As previous chapters have docu-
mented in detail, the targeting process was inefficient to a fault,
command and control arrangements were excessively complicated,
and enemy IADS challenges indicated much unfinished work for
SEAD planners. In addition, elusive enemy ground forces belied the
oft-cited claim of airmen that air power has arrived at the threshold
of being able to find, fix, track, target, and engage any object on the
surface of the earth. 32
The biggest failures of Allied Force likewise occurred in the realm of

strategy and execution. First, despite its successful outcome and
through no fault of allied airmen, the bombing effort was clearly a
suboptimal application of air power. The incremental plan chosen
by NATO’s leaders risked squandering much of the capital that had
been built up in air power’s account ever since its ringing success in
______________
31Interview with Air Marshal Sir John Day, RAF, UK Ministry of Defense director of
32These and other surprises should stand as a sobering reminder that the compara-
tively seamless and unfettered successes achieved by allied air power during Opera-
tion Desert Storm were most likely the exception rather than the rule for future joint
and combined operations—both the operating area and the circumstances surround-
ing the 1991 Gulf War were unique. For more on this point, see Air Vice Marshal Tony
Mason, RAF (Ret.), Air Power: A Centennial Appraisal, London, Brassey’s, 1994, pp.
140–158.
Desert Storm nearly a decade before. General Clark’s early comment

that NATO would “grind away” at Milosevic rather than hammer him
hard and with determination attested powerfully to the watered-
down nature of the strikes.33 By meting out those strikes with such
hesitancy, NATO’s leaders remained blind to the fact that air power’s
very strengths can become weaknesses if the air weapon is used in a
way that undermines its credibility.34 Almost without question, the
first month of underachievement in the air war convinced Milosevic
that he could ride out the NATO assault.
Indeed, the way Operation Allied Force commenced violated two of

the most enduring axioms of military practice: the importance of
achieving surprise and the criticality of keeping the enemy unclear as
to one’s intentions. The acceptance by NATO’s leaders of a strategy
that preemptively ruled out a ground threat and envisaged only
gradually escalating air strikes to inflict pain was a guaranteed recipe
for downstream trouble, even though it was the only strategy that, at
the time, seemed politically workable. For U.S. defense leaders to
have suggested afterward that NATO’s attacks against fielded enemy
ground troops “forced [those troops] to remain largely hidden from
view . . . and made them ineffective as a tactical maneuver force” and
that its SEAD operations forced Milosevic to “husband his antiair-
craft missile defenses to sustain his challenge [to NATO air opera-
tions]” was to make a virtue of necessity on two counts.35 First, it
was the absence of a credible NATO ground threat that enabled Milo-
sevic’s troops to disperse and hide, making it that much more diffi-
cult for NATO’s aircrews to find and attack them. The ineffectiveness
of those troops as a tactical maneuver force was quite beside the
point, considering that tactical maneuver was not required for the
ethnic cleansing those troops managed to sustain quite handily
throughout most of the air war’s duration. Second, it would have
been more honest to say that the Serb tactic of carefully conserving
antiaircraft missile defenses throughout Allied Force made those
______________
33Eric Schmitt, “Weak Serb Defense Puzzles NATO,” New York Times, March 26, 1999.
34For a fuller development of this point, see Daniel L. Byman, Matthew C. Waxman,
and Eric Larson, Air Power as a Coercive Instrument , Santa Monica, California, RAND,
MR-1061-AF, 1999. See also Grant T. Hammond, “Myths of the Air War Over Serbia:
Some ‘Lessons’ Not to Learn,” Aerospace Power Journal, Winter 2000, pp. 78–86.
35Cohen and Shelton, “Joint Statement.”
defenses a continuing threat to NATO’s freedom to operate in Yu-

goslav airspace, undermining the effectiveness of many sorties as a
result.
In fairness to the U.S. and NATO officials most responsible for air op-
erations planning, many of the differences between Allied Force and
the more satisfying Desert Storm experience were beyond the control
of the allies, and they should be duly noted in any critique of the way
the former was conducted. To begin with, as discussed earlier, bad
weather was the rule, not the exception. Second, variegated and
forested terrain limited the effectiveness of many sensors. Third,
Serb SAM operators were more proficient and tactically astute than
those of Iraq. Fourth, alliance complications were greater by far in
Allied Force than were the largely inconsequential intracoalition dif-
ferences during the Gulf War. Finally, because the goal of Allied
Force was more to compel than to destroy, it was naturally more dif-
ficult for senior decisionmakers to measure and assess the air war’s
daily progress, since there was no feedback mechanism to indicate
how well the bombing was advancing toward coercing Milosevic to
comply with NATO’s demands. It was largely for that reason that
most Allied Force planners were surprised when he finally decided to
capitulate.
That said, the most important question with respect to Allied Force
has to do less with platform or systems performance than with the
more basic strategy choices that NATO’s leaders made and what
those choices may suggest about earlier lessons forgotten—not only
from Desert Storm and Deliberate Force but also from Vietnam. Had
Milosevic been content to hunker down and wait out NATO’s
bombing effort, he could easily have challenged the long-term cohe-
sion and staying power of the alliance. Fortunately for the success of
Allied Force, by opting instead to accelerate his ethnic cleansing of
Kosovo, he not only united the West in revulsion but also left NATO
with no alternative but to dig in for the long haul, both to secure an
outcome that would enable the repatriation of nearly a million dis-
placed Kosovars and to ensure its continued credibility as a military
alliance.
Because of the almost universal assumption among NATO’s leaders

that the operation would last no more than two to four days, the first
30 days of the air war were badly underresourced. Among the results
of this erroneous assumption were erratic procedures for target

nomination and review, too few combat aircraft on hand for con-
ducting both night and day operations, and pressures from SACEUR
for simultaneous attacks not only on fixed infrastructure targets but
also on fielded VJ forces. Relatedly, there was an inadequate airspace
management plan and no flexible targeting cell in the CAOC for ser-
vicing SACEUR’s sudden demands to attack VJ forces in the KEZ. All
of these problems, it bears stressing, were a reflection not on NATO’s
air power or its mechanisms for using air power per se, but rather on
the strategy choices that were made (or, perhaps more correctly, for-
gone) by NATO’s political leaders.
To be sure, allied capabilities for detecting and engaging fleeting en-

emy ground-force targets improved perceptibly as the weather grew
more agreeable with approaching summer and as the KLA became
more active. Nevertheless, persistent problems with the flexible tar-
geting effort spotlighted further work that needs to be done. The
CAOC went into the operation without an on-hand cadre of experi-
enced target planners accustomed to working together harmo-
niously. Accordingly, General Short was forced to resort to a “pick-
up team” during the first month of operations against VJ forces in
Kosovo. The fusion cell also frequently lacked ready access to all-
source reconnaissance information. At first, data from special oper-
ations forces and the Army’s TPQ-36 and TPQ-37 firefinder radars in
Albania were not provided to the CAOC. Indeed, there was an ab-
sence of allied ground-force representation in the CAOC until the air
war’s very end. Other needs that became apparent included regular-
ized and centralized mensuration of target coordinates as new target
candidates were detected and became available for prompt servicing.
Beyond that, the very nature of Operation Allied Force and the man-
ner in which it was conducted from the highest levels both in Wash-
ington and in Brussels placed unique stresses on the JFACC’s ability
to command and control allied air operations. For example, General
Short and his staff had to contend on an unrelenting basis with rapid
shifts in political priorities and SACEUR guidance, as well as with the
myriad pressures occasioned by a random and nonsystematic flow of
assets to the theater, ranging from combat aircraft to staff aug-
mentees in the CAOC. All of these problems emanated from a lack of
consensus among the top decisionmakers on both sides of the At-
lantic as to what the air effort’s military goals were at any given mo-
ment and what it would take to “prevail.” The de facto “no friendly
loss” rule, stringent collateral damage constraints, and the absence
of a NATO ground threat to force VJ troops to concentrate and thus
make them easier targets further limited the rational employment of
available in-theater assets and placed a premium on accurate infor-
mation and the use of measures that took a disconcertingly long time
to plan, carry out, and evaluate. 36 One realization driven home by
these and other shortcomings was the need for planners in the tar-
geting cell to train together routinely in peacetime before a contin-
gency requires them to react at peak efficiency from the very start.
GRADUALISM AND ITS IMPLICATIONS

The greatest frustration of Operation Allied Force was its slow start
and equally slow escalation. A close second entailed the uniquely
stringent rules of engagement that limited the effectiveness of many
combat sorties. Indeed, the dominance of political inhibitions was a
signal feature of the air war from start to finish. Because it was an
operation performed essentially for humanitarian purposes, neither
the United States nor any of the European members of NATO saw
their security interests threatened by ongoing events in Yugoslavia.
Because the perceived stakes were not high, at least at the outset, any
early commitment by NATO to a ground offensive was all but out of
the question. Moreover, both the anticipated length of the bombing
effort and the menu of targets attacked were bound to be matters of
often heated contention.
On top of that, the avoidance of noncombatant fatalities among

Yugoslavia’s civilian population was rightly of paramount concern
to NATO’s leaders, further aggravating the complications caused
by poor target-area weather throughout much of the air war. As
USEUCOM’s director of operations, USAF Major General Ronald
Keys, later noted, while there was no single target whose elimination
might have won the war, there was a profusion of targets that could
potentially have lost the war had they been struck, either intention-
ally or inadvertently. In the presence of factors like these that could
have split the alliance at any time, NATO’s unity was a sine qua non
______________
36 I am indebted to my RAND colleagues James Schneider, Myron Hura, and Gary
McLeod for these on-target summary observations.
for the success of Allied Force. Not surprisingly, the Serbs were
aware of that fact and were frequently able to exploit it.37
Acceptance of these realities, however, hardly eased the discomfiture

among air warfare professionals over the fact that NATO’s self-
imposed restraints were forcing them to fight with one hand tied be-
hind their backs. One analyst, reporting the results of interviews
conducted in late April with some two dozen senior active and re-
tired Air Force generals, reported a collective sense of “disappoint-
ment that air power is being so poorly employed [and] frustration
over the false promise of a perfect war and zero casualties.” His in-
terviews revealed a deep-seated concern that “with far too much po-
litical micromanagement but without a clear strategy and the aid of
ground forces, the air war . . . is destined to fail.” Worst of all, the
generals complained, the United States and NATO did not take ad-
vantage of the shock effect of air power. Said retired General Charles
Horner: “We are training [the Serbs] to live with air attacks.” Said
another Air Force general: “Air planners are not planning the air op-
eration. They are being issued targets each day for the next day’s op-
erations, too late to do rational planning.”38
There was no less disaffection among air warfare professionals at the

working level. As one U.S. pilot flying combat sorties complained in
an email message that made its way to public light: “This has been a
farce from the start. We have violated every principle of campaign
air power I can think of.” The pilot hastened to add that “over-
zealous air power advocates have, since Desert Storm, sold us as
something we are not. Air power can do a lot of things, [but] it
cannot change the mind of a dictator who has his people’s tacit sup-
port.” Nevertheless, he concluded, “it is not the USAF’s fault that the
air campaign is not going as well as Desert Storm. Hitting 5–8 targets
a night, with sequential [as opposed to] parallel operations, is not the
way to prosecute a campaign.”39
______________
37Cited in Colonel Steve Pitotti, USAF, “Global Environments, Threats, and Military
Strategy (GETM) Update,” Air Armament Summit 2000 briefing, 2000.
38William M. Arkin, “Inside the Air Force, Officers Are Frustrated About the Air War,”
39Rowan Scarborough, “Officers Criticize Air War Strategy,” Washington Times, May
10, 1999.
The UK Ministry of Defense’s director of operations in Allied Force

denied that there was ever a hard-and-fast rule that NATO must not
lose an aircraft under any circumstances.40 Yet NATO’s leaders had
powerful incentives to avoid any circumstances that might result in
friendly aircrews being killed in action or taken prisoner of war, since
the continued cohesion of the alliance was the latter’s center of grav-
ity and since any such losses would have been precisely the sorts of
untoward events most likely to undermine it. Indeed, if there was
any unwritten “prime directive” that guided NATO’s strategy
throughout the course of Allied Force, it was the preservation of its
own solidity, especially during the air war’s critical early weeks. In
light of that concern, General Short admitted toward the end of May
that zero losses was a primary goal in fact if not in name.41 Not only
would a split in the alliance have undermined the air war’s effort
against Belgrade, it would have raised fundamental questions about
the future viability of NATO as a military alliance. It naturally fol-
lowed that an incremental bombing effort and least-common-
denominator targeting had to be accepted until it became clearer
throughout the alliance that NATO was committed for the long
haul.42
______________
40Interview with Air Marshal Sir John Day, RAF, UK Ministry of Defense director of
41William Drozdiak, “Air War Commander Says Kosovo Victory Near,” Washington
Post, May 24, 1999. Clark himself later indicated that his chief “measure of merit” in
keeping Allied Force on track was “not to lose aircraft, minimize the loss of aircraft.”
He further stated that this exacting desideratum “drove our decisions on tactics, tar-
gets, and which airplanes could participate,” but that it was motivated by a “larger po-
litical rationale: if we wanted to keep this campaign going indefinitely, we had to pro-
tect our air fleet. Nothing would hurt us more with public opinion than headlines that
screamed, ‘NATO LOSES TEN AIRPLANES IN TWO DAYS.’” General Wesley K. Clark,
Waging Modern War: Bosnia, Kosovo, and the Future of Combat, New York, Public
Affairs, 2001, p. 183.
42It bears noting that the zero-loss issue, however seriously it may have been regarded
at the highest leadership levels, had little day-to-day impact on actual combat opera-
tions. As an F-15E instructor Electronic Warfare Officer (EWO) who flew multiple
combat missions with the 494th Fighter Squadron recalled from first-hand experience:
“The issue of the ‘no-losses rule’ did not filter down to the aircrew level, since we al-
ways plan with that goal in mind. We were briefed that there were no ‘high-priority’
targets prior to the opening of hostilities, but that ended up having little effect on the
risk level that we were willing to accept. The concrete effects of the ‘no-loss rule’ were
the 15,000-ft floor and a number of unreasonable ROE restrictions. However, outside
the immediate tactical constraints imposed by the ROE, the prevailing high-level atti-
tude had no effect on tactical operations. We were aware of the priority placed on
Although the manner in which Allied Force was conducted fell short
of the ideal use of air power, it suggests that gradualism may be here
to stay if U.S. leaders ever again intend to fight wars for marginal or
amorphous interests with as disparate a set of allies as NATO. As the
vice chairman of the JCS at the time, USAF General Joseph Ralston,
noted after the air effort ended, air warfare professionals will con-
tinue to insist, and rightly so, that a massive application of air power
will be more effective than gradualism. Yet, Ralston added, “when
the political and tactical constraints imposed on air use are extensive
and pervasive—and that trend seems more rather than less likely—
then gradualism may be perceived as the only option.”43 General
Jumper likewise intimated that the United States may have little
choice but to accept the burdens of an incremental approach as an
unavoidable cost of working with shaky allies and domestic support
in the future: “It is the politics of the moment that will dictate what
we can do. . . . If the limits of that consensus mean gradualism, then
we’re going to have to find a way to deal with a phased air campaign.
Efficiency may be second.” 44
Insofar as gradualism promises to be the wave of the future, it sug-

gests that airmen will need to discipline their natural urge to bridle
whenever politicians hamper the application of a doctrinally pure
campaign strategy and to recognize and accept instead that political
considerations, after all, determine—or should determine—the way
in which campaigns and wars are fought. This does not mean that
military leaders should surrender to political pressures without first
making their best case for using force in the most effective and cost-
minimizing way. It does, however, stand as an important reminder
_____________________________________________________________
minimizing losses, but the effect on the mission was overrated. There were no cases
that I am aware of where the aircrew said, ‘Well, this looks a little hairy, and the prior-
ity is not to lose an airplane, so I won’t do it.’ We were more likely to abort an attack
for collateral damage concerns than we were to abort for survivability issues. As would
be expected, aircrews pressed to the target in the face of serious opposition.” Major
Michael Pietrucha, USAF, personal communication to the author, July 9, 2001.
43“Ralston Sees Potential for More Wars of Gradual Escalation,” Inside the Pentagon,
September 16, 1999, p. 1.
44“Washington Outlook,” Aviation Week and Space Technology, August 23, 1999, p. 27.
It hardly follows, of course, that gradualism and coalitions must invariably be syn-
onymous. They certainly were not in Desert Storm in 1991. Clearly, the extent to
which gradualist strategies will prove unavoidable in the future will depend heavily on
both the shared stakes for would-be coalition partners and the skill of their leaders in
setting the direction and tone of coalition conduct.
that war is ultimately about politics and that civilian control of the
military is an inherent part of the democratic tradition. It follows
that although airmen and other warfighters are duty-bound to try to
persuade their civilian superiors of the merits of their recommenda-
tions, they also have a duty to live with the hands they are dealt and
to bend every effort to make the most of them in an imperfect
world.45 It also follows that civilian leaders at the highest levels have
an equal obligation to try to stack the deck in such a manner that the
military has the best possible hand to play and the fullest possible
freedom to play it to the best of its ability. This means expending the
energy and political capital needed to develop and enforce a strategy
that maximizes the probability of military success. In Allied Force,
that was not done by the vast majority of the top civilian leaders on
either side of the Atlantic. 46
On the plus side, the air war’s successful outcome despite its many
frustrations suggested that U.S. air power may now have become ca-
pable enough, at least in some circumstances, to underwrite a strat-
egy of incremental escalation irrespective of the latter’s inherent
inefficiencies. What made the gradualism of Allied Force more bear-
able than that of the earlier war in Vietnam is that NATO’s advan-
tages in stealth, precision standoff attack, and electronic warfare
meant that it could fight a one-sided war against Milosevic with
near-impunity and achieve the desired result, even if not in the most
ideal way. 47 That was not an option when U.S. air power was a less
developed tool than it is today.
______________
45On this point, Air Vice Marshal Mason remarked that he had not “spent the past 25
years trying to persuade unbelievers of the efficacy of air power only to finish up
whining because political circumstances made operations difficult.” Personal com-
munication to the author, October 22, 1999. In a similar spirit, the leader of USAFE’s
post–Allied Force munitions effectiveness investigation in Kosovo later suggested that
airmen should “consider a politically restricted target list like the weather: complain
about it, but deal with it.” Colonel Brian McDonald, USAF, briefing at RAND, Santa
Monica, California, December 14, 1999.
46It further follows that airmen, for their part, need to learn not only how to conduct
gradual campaigns more effectively, but also how better to explain convincingly to
politicians the value of using mass and shock early and the greater strategic effective-
ness of effects-based targeting.
47See Colonel Phillip S. Meilinger, USAF, “Gradual Escalation: NATO’s Kosovo Air
Campaign, Though Decried as a Strategy, May Be the Future of War,” Armed Forces
Journal International, October 1999, p. 18.
On this point, Admiral Ellis, a career fighter pilot himself, was no less
disturbed by the air war’s lethargic pace than was his air component
commander, General Short, or any other airmen on down the line.
However, mindful of the long-standing political and bureaucratic
rule of thumb that “if a problem has no solution, it is no longer a
problem but a fact,” he recognized that ideal-world solutions were
unworkable in the Allied Force setting and that flexibility was re-
quired in applying air doctrine in a difficult situation. As it turned
out, NATO conducted its bombing effort in a way that was not max-
imally efficient, yet that worked in the end to foil Serb strategy, which
was to wait out the alliance and strive mightily to fragment it. Be-
cause the escalation was gradual over time, the coalition succeeded
in holding together. Because NATO used highly conservative tactics,
it lost no aircrews and civilian casualties and collateral damage were
kept to a minimum. In effect, a compromise was struck in which the
air war was intense enough to maintain constant pressure on Milo-
sevic yet measured enough to keep NATO from falling apart. Either
the loss of friendly lives beyond token numbers or an especially grue-
some spectacle of collateral damage could have been more than
enough to incline at least some key allies to call it quits. Noting fur-
ther that NATO fought in this case to establish conditions rather than
to “win” in the classic sense, Ellis added that a campaign strategy
that would have allowed Desert Storm–like intensity and scale of tar-
get attacks to be employed was simply never in the cards.
By the same token, RAF Air Commodore Andrew Vallance pointed

out that because a key attraction of air power to civilian decision-
makers is its adaptability for accommodating different situations in
different ways as needed, “the purist ‘one size fits all’ approach to air
doctrine needs to be moderated. Existing air doctrine is fine for
high-intensity conflicts, but more subtle operational doctrines are
needed in the complex world of peace support.” 48 Echoing this
point, Karl Mueller observed that “sometimes strategists will be
called upon to execute gradually escalatory air campaigns whether
they approve of the concept or not, and thus they should develop
some expertise in the art form even if they abhor it.”49 With the air
______________
48Air Commodore A. G. B. Vallance, RAF, “After Kosovo: Implications of Operation Al-
lied Force for Air Power Development,” unpublished paper, p. 4.
49Mueller, “Deus ex Machina?” p. 16.
weapon now largely perfected for such canonical situations as halt-

ing massed armored assaults, it needs to be further refined for han-
dling messier, less predictable, and more challenging combat situa-
tions featuring elusive or hidden enemy ground forces, restrictive
rules of engagement, disagreeable weather, the enemy use of human
shields, lawyers in the targeting loop as a matter of standard practice,
and diverse allies with their own political agendas, all of which were
characteristic features of the Kosovo crisis. Moreover, although
NATO’s political leaders arguably set the bar too high with unrealistic
expectations about collateral damage avoidance, it seems clear that
the Western democracies have long since passed the point where
they can contemplate using air power, or any force, for that matter,
in as unrestrained a way as was characteristic of World War II
bombing. Admiral Ellis noted that NATO barely averted legal conse-
quences prompted by the collateral damage incidents that occurred
in Allied Force. This implies that along with new precision-attack
capability goes new responsibility, and air warfare professionals
must now understand that they will be held accountable.50
On this point, one can fairly suggest that both SACEUR and his
JFACC were equally prone throughout Allied Force to remain wedded
to excessively parochial views of their preferred target priorities,
based on implicit faith in the inherent correctness of their respective
services’ doctrinal teachings. They might more effectively have ap-
proached Milosevic instead as a unique rather than generic oppo-
nent, conducted a serious analysis of his distinctive vulnerabilities,
and then tailored a campaign plan aimed at attacking those vulner-
abilities directly, irrespective of canonical air or land warfare solu-
tions for all seasons. A year after the air war, in a measured reflection
on the recurrent tension that afflicted the interaction of Clark and
Short, Admiral Ellis suggested that the failure of all the services to
advance beyond their propensity to teach only pristine, service-ori-
ented doctrines at their respective war colleges reflected a serious
“cultures” problem and that the services badly need to plan for and
______________
50This includes being held increasingly accountable for their own combat losses. The
Allied Force SEAD experience showed that in crises where less-than-vital U.S. interests
are at stake, near-zero attrition of friendly aircraft and their aircrews will be a high, and
possibly determining, priority governing operational tactics.
accommodate the unexpected and the unconventional, both of

which were daily facts of life during Operation Allied Force. 51
Finally, the probability that coalition operations in the future will be

the rule rather than the exception suggests a need, to the fullest ex-
tent practicable, to work out basic ground rules before a campaign
begins, so that operators, once empowered, can implement the
agreed-upon plan with a minimum of political friction. As it was,
Allied Force attested not only to the strategy legitimation that comes
from the force of numbers provided by working through a coalition,
but also to the limitations of committee planning and least-
common-denominator targeting. General Short commented that
the need for 19 approvals of target nominations was “counterpro-
ductive” and that an appropriate conclusion was that “before you
drop the first bomb or fire the first shot, we need to lock the political
leaders up in a room and have them decide what the rules of en-
gagement will be so they can provide the military with the proper
guidance and latitude needed to prosecute the war.”52 As it was,
Short later said in his PBS interview, the rules continuously ebbed
and flowed in reaction to events over the air war’s 78 days: “You can
go to downtown Belgrade, oh my God, you’ve hit the Chinese em-
bassy, now there’s a five-mile circle going around downtown Bel-
grade into which you cannot go.” As a result, he complained, strikers
______________
51Conversation with Admiral James Ellis, USN, Headquarters Allied Forces Southern
Europe, Naples, Italy, May 30, 2000.
52William Drozdiak, “Allies Need Upgrade, General Says.” As sensible as this sugges-
tion may have sounded after the fact, however, one must ask how workable it would
have been in actual practice. Wars characteristically feature dynamics that push par-
ticipants beyond anything imaginable at the outset. Setting clear going-in rules is easy
and feasible enough for something short and relatively straightforward, like Operation
Deliberate Force and Operation El Dorado Canyon, the joint USAF-Navy raid on Libya
in 1986. Expecting them in larger and more open-ended operations, however, means
counting on a predictability of events that does not exist in real life. The fact is that
there was a consensus at the start of Allied Force about what was acceptable and what
everyone was willing to do, and that was for 91 targets and two nights of bombing.
NATO’s cardinal error was not its failure to reach a consensus before firing the first
shot; it was its refusal to be honest up front about what it would do if its assumptions
about Milosevic’s resolve proved false. I thank Dr. Daniel Harrington, Office of
History, Hq USAFE, for having shared this insightful observation with me. I would add
that had NATO’s leaders done better at attending to that responsibility, they would
have gone a long way toward satisfying General Short’s expressed concern.
often ended up “bombing fire hydrants and stoplights because there

just weren’t targets of great value left that weren’t in a sanctuary.”53
THE COST OF THE MISSING GROUND THREAT

One of the most important realizations to emerge from Allied Force
at the operational and strategic levels was that a ground component
to joint campaign strategies may be essential, at least in some cases,
for enabling air power to deliver to its fullest potential. The com-
mander of Air Combat Command, General Richard Hawley, was one
of many senior airmen who freely admitted that the a priori decision
by the Clinton administration and NATO’s political leaders not to
employ ground forces had undercut the effectiveness of allied air op-
erations: “When you don’t have that synergy, things take longer and
they’re harder, and that’s what you’re seeing in this conflict.”54
General Jumper later concluded similarly that the imperative of at-

tacking fielded enemy forces without the shaping presence of a
NATO ground threat had produced “major challenges,” including
creating a faster flexible targeting cycle; putting a laser designator on
Predator; creating new target development processes within the
CAOC; creating real-time communications links between finders, as-
sessors, and shooters; and developing more rapid real-time retarget-
ing procedures for the B-2s, the B-1s, the B-52s, and F-15Es carrying
the AGM-130.55
Amplifying on the fallacy of having started the air effort without a

credible ground threat, General Short noted that “this conflict was
unlike others in that we did not have a ground element to fix the en-
emy, to make him predictable, and to give us information as to where
the enemy might be.”56 Short went on to point out, however, that
although NATO had not been formally allied with the KLA, the fact
______________
rope,” February 22, 2000.
54Bradley Graham, “General Says U.S. Readiness Is Ailing,” Washington Post, April 30,
1999.
56Drozdiak, “Allies Need Upgrade, General Says.”
that the latter had begun to operate with some success in the end
“made the Yugoslav army come out and fight and try to blunt their
offensive. . . . And once they moved, or fired their artillery, our
strikers learned where they were and could go in for the kill.”57 Had
VJ forces in Kosovo faced an imminent NATO ground invasion, or
even a credible threat of such an invasion later, they would have
been obliged to move troops and supplies over bridges that NATO
aircraft could have dropped. They also would have been compelled
to concentrate and maneuver in ways that would have made it easier
for NATO to find and attack them.
Earlier, White House national security adviser Samuel Berger main-

tained that taking ground forces off the table at the outset had been
the right thing to do because anything else would have inevitably
prompted an immediate public debate both in the United States and
among the allies, which could have split the alliance and seriously
impeded the overall air effort.58 Yet there was a huge difference be-
tween acknowledging that a land offensive could be fraught with
danger, on the one hand, and ruling out such an offensive categori-
cally before the fact, on the other. The former would have been de-
manding enough even under the best of circumstances because of
basing, airlift, and logistics problems. The latter, however, was a
colossal strategic mistake, in that it gave Milosevic the freedom to act
against the Kosovar Albanians and the power to determine when the
war would be over. The opportunity costs incurred by NATO’s ane-
mic start of Allied Force without an accompanying ground threat in-
cluded a failure to exploit air power’s shock potential and to instill in
Milosevic an early fear of worse consequences yet to come; the en-
couragement it gave VJ troops to disperse and hide while they had
time; the virtual carte blanche it gave Milosevic for accelerated
atrocities in Kosovo; and the relinquishment of the power of initia-
tive to the enemy.
As for the oft-noted concern over the prospect of sustaining an un-

bearable level of friendly casualties had NATO opted to back up the
air war with a ground element, there most likely would have been no
______________
57Ibid.
les Times, June 9, 2000.
need actually to commit NATO troops to battle in the end. The mere
fact of a serious Desert Shield–like deployment of NATO ground
troops along the Albanian and Macedonian borders would have
made their VJ counterparts more easily targetable by allied air power.
Had such a deployment commenced in earnest, it also might have
helped to deter, or at least lessen, the ethnic cleansing of Kosovo by
giving VJ troops a more serious concern to worry about. In both
cases, it could have enabled a quicker end to the war.
Even had Milosevic remained unyielding to the point where an op-

posed NATO ground-force entry would have been unavoidable
sooner or later, continued air preparation of the battlefield might
have been sufficiently effective that the VJ’s residual strength would
not have presented a significant challenge to NATO land forces. The
impending improvement of summer weather and the further estab-
lishment of NATO air dominance would have enabled more effective
NATO air performance against VJ targets, especially had the KLA suc-
ceeded in maintaining enough pressure to force VJ units to bunch up
and move.
Indeed, well before Allied Force ended, there was a gathering sense
among some observers that Serbia’s ground forces were being given
more credit than they deserved as an excuse for ruling out a NATO
land-invasion option. As one former U.S. Army officer pointed out,
Milosevic’s army was a small conscript-based force with an active
component of only some 115,000 troops who relied on antiquated
Soviet equipment, mainly the 1950s-vintage T-55 tank. Air strikes
during the first few nights of Allied Force had already rendered Yu-
goslavia’s small air force a non-factor in any potential NATO ground
push. The VJ’s petroleum and other stocks for sustainment had also
been rapidly depleted by the bombing, leaving the Serbs with, at
best, only a minimal capacity to wage conventional war against a se-
rious ground opponent. In contrast, the modern and well-equipped
NATO ground forces arguably possessed enough combat power “to
make mincemeat of the Yugoslav army.”59
Be that as it may, the problems created by NATO’s having ruled out a

ground option before the fact suggest an important corrective to the
______________
59Andrew J. Bacevich, “Target Belgrade: Why a Ground War Would Be a Rout,” Na-
tional Review, May 3, 1999, p. 29.
seemingly unending argument between airmen and land combatants

over the relative merits of air power versus “boots on the ground.”
Although Operation Allied Force reconfirmed that friendly ground
forces no longer need to be inexorably committed to combat early, it
also reconfirmed that air power in many cases cannot perform to its
fullest potential without the presence of a credible ground
component in the campaign strategy. The fact is that air power alone
was not well suited to defeating VJ forces in the field. Once most of
the combat returns were in, it became clear that few allied kills were
accomplished against dispersed and hidden VJ units in the KEZ. Not
only that, allied air power had been unable to protect the Kosovar Al-
banians from Serb terror tactics, a problem that was further exacer-
bated by the stringent rules of engagement aimed at minimizing
collateral damage and avoiding any NATO loss of life. As former Air
Force chief of staff General Merrill McPeak instructively elaborated
on this point, “in a major blunder, the use of ground troops was ruled
out from the beginning. I know of no airman—not a single one—
who welcomed this development. Nobody said, ‘Hey, finally, our
own private war. Just what we’ve always wanted!’ It certainly would
have been smarter to retain all the options. . . . Signaling to Belgrade
our extreme reluctance to fight on the ground made it much less
likely that the bombing would succeed, exploring the limits of air
power as a military and diplomatic instrument.”60
TOWARD A “REPORT CARD” FOR ALLIED FORCE

As for what airmen and other observers should take away from Allied
Force by way of lessons indicated and points worth pondering, the
commander of the U.S. military contribution, Admiral Ellis, offered a
good start when he declared in his after-action briefing to Pentagon
and NATO officials that luck played the chief role in ensuring the air
______________
60General Merrill A. McPeak, USAF (Ret.), “The Kosovo Result: The Facts Speak for
Themselves,” Armed Forces Journal International, September 1999, p. 64. In a similar
vein, the chief of staff of the RAF later faulted NATO’s decision to rule out a ground
option from the start of the air war as “a strategic mistake” that enabled Serb forces to
forgo preparing defensive positions, hide their tanks and artillery and make maximum
use of deception against NATO attack efforts, and conduct their ethnic cleansing of
Kosovo with impunity. Michael Evans, “Ground War ‘Error,’” London Times, March
24, 2000.
war’s success.61 Ellis charged that NATO’s leaders “called this one
absolutely wrong” and that their failure to anticipate what might oc-
cur once their initial strategy of hope failed occasioned most of the
untoward consequences that ensued thereafter.62 These included
the hasty activation of a joint task force, a race to find suitable tar-
gets, an absence of coherent campaign planning, and lost opportu-
nities caused by the failure to think through unpleasant excursions
from what had been expected. Ellis concluded that the imperatives
of consensus politics within NATO made for an “incremental war”
rather than for “decisive operations,” that excessive concern over
collateral damage created “sanctuaries and opportunities for the
adversary—which were successfully exploited,” and that the lack of a
credible NATO ground threat “probably prolonged the air cam-
paign.”63 It was only because Milosevic made a blunder no less tow-
ering than NATO’s preclusion of a ground option that the war had
the largely positive outcome that it did.
Indeed, that NATO prevailed in the end with only two aircraft lost
and no combat fatalities sustained surely reflected good fortune at
least as much as the professionalism of its aircrews and their com-
manders. General Jumper explained afterward that “we set the bar
fairly high when we fly more than 30,000 combat sorties and we don’t
lose one pilot. It makes it look as if air power is indeed risk free and
too easy a choice to make.” Amplifying on the same point, retired
RAF Air Vice Marshal Tony Mason observed that seeking to minimize
one’s losses is both admirable and proper up to a point, yet it can
lead to self-deterrence when efforts to escape the costs of war are
______________
61Amplifying on his suggestion that luck was the key player, Ellis pointed out how
much worse matters would have been for the alliance had NATO experienced any one
of a number of untoward developments: an enemy attack on its troops deployed in
theater with ground forces or tactical ballistic missiles; the possibility of even a few
NATO aircrews being killed in action or captured as POWs; the continuation of the
fighting into the winter; the depletion of U.S. precision munitions stocks; the weaken-
ing or evaporation of public support; an allied ground invasion becoming the only op-
tion; or a decision by France or Italy to withdraw from further participation.
62Revealingly, barely a week into Allied Force, one senior Clinton administration offi-
cial, when asked what NATO’s strategy would be should Phase III of the air war fail to
persuade Milosevic to admit defeat, replied: “There is no Phase IV.” Quoted in John
Broder, “In Grim Week, Pep Talk from the President,” New York Times, April 1, 1999.
63Elaine M. Grossman, “For U.S. Commander in Kosovo, Luck Played Role in Wartime
Success,” Inside the Pentagon, September 9, 1999, p. 1.
pursued to a moral fault. Although force protection “must be a major

concern for any force commander,” Mason added, “my own view is
that if Saint George’s first priority with tackling dragons had been
force protection, I don’t think he would now be the patron saint of
England.”64
The Kosovo experience further suggested some needed changes in

both investment strategy and campaign planning. The combination
of marginal weather and the unprecedented stress placed on avoid-
ing collateral damage made for numerous days between March 24
and mid-May when entire ATOs had to be canceled and when only
cruise missiles and the B-2, with its through-the-weather JDAM ca-
pability, could be used. That spoke powerfully for broadening the
ability of other aircraft to deliver accurate munitions irrespective of
weather, as well as for ensuring that adequate stocks of such muni-
tions are on hand to see the next campaign to completion. The ex-
tended stretch of bad weather underscored the limitations of LGBs
and confirmed the value of GPS-guided weapons like JDAM that can
bomb accurately through the weather.
Not surprisingly, the munitions used in Allied Force generally per-

formed as advertised. The operation’s results, however, confirmed
the need for a larger U.S. inventory of precision-guided munitions
(especially those capable of all-weather target attack), as well as
greater accuracy and more standoff attack capability. At the same
time, it indicated a continued operational utility for both unguided
general-purpose bombs and cluster munitions for engaging soft mili-
tary area targets deployed in the open. Other areas in which allied
weapons performance showed a need for further improvement in-
clude interoperability across platforms, more multispectral sensors,
higher-gain optical sensors for UAVs, more data-link interoperability,
a wider range of bomb sizes, and weapons capable of conducting
“auto-BDA.”65 Still other force capability needs highlighted by the
______________
64Comments at an Air Force Association Eaker Institute colloquy, “Operation Allied
Force: Strategy, Execution, Implications,” held at the Ronald Reagan International
Trade Center, Washington, D.C., August 16, 1999.
65These were among numerous other conclusions suggested by Major General Ronald
Keys, USAF, director of operations (J-3), U.S. European Command, cited in Colonel
Steve Pitotti, USAF, “Global Environments, Threats, and Military Strategy (GETM) Up-
date,” Air Armament Summit 2000 briefing, Eglin AFB, Florida, 2000.
Allied Force experience include better means for locating moving

targets, better discrimination of real targets from decoys, and a way
of engaging those targets with smart submunitions rather than with
more-costly PGMs and cruise missiles. 66 One airman later com-
mented frankly that in being tasked by Clark to go after dispersed
and hidden VJ forces, U.S. air power “was being asked to be a 21st
century tactical air force . . . and the truth is, we’re not very good at
it,” at least yet.67
As for the ultimate wisdom of the allied decision to proceed with the
air war in the first place, the United States and NATO displayed
an ability in this case to apply coercion successfully through air
power from a poorly prepared battlefield at a remarkably low cost in
noncombatant fatalities caused by direct collateral damage.68 Yet
there is a danger that making a habit of such displays by accepting
Allied Force as a model for future interventions could easily lead to
an erosion of the U.S. claim to global leadership.69 On the contrary,
Allied Force should have underscored the fact that one of the most
acute challenges facing U.S. policymakers in the age of a single su-
perpower entails deciding when, and in what manner, to intervene in
humanitarian crises that do not yet impinge directly on U.S. security
interests.
ON THE USES AND ABUSES OF AIR POWER

Viewed in hindsight, the most remarkable thing about Operation Al-
lied Force was not that it defeated Milosevic in the end, but rather
______________
66Work on this is being performed by Alan Vick of RAND.
67Elaine M. Grossman, “U.S. Military Debates Link Between Kosovo Air War, Stated
Objectives,” Inside the Pentagon, April 20, 2000, p. 6.
68A heated argument arose after the war ended between defenders and critics of the
Clinton administration’s strategy for Kosovo over whether the approach taken, despite
its low cost in noncombatant lives lost to direct collateral damage, nonetheless
produced an unconscionably high loss of civilian innocents to the Serbian ethnic
cleansing campaign which it allegedly accelerated. For a snapshot summary of the
positions taken on both sides, see Christopher Layne and Benjamin Schwarz, “Kosovo
II: For the Record,” The National Interest, Fall 1999, pp. 9–15, and Ivo Daalder, “NATO
and Kosovo,” The National Interest, Winter 1999/2000, pp. 113–117.
69I am grateful to Lieutenant General Bradley Hosmer, USAF (Ret.), for bringing this
point to my attention.
that air power prevailed despite a U.S. leadership that was unwilling
to take major risks and an alliance that held together only with often
paralyzing drag. Fortunately, the Clinton administration did a cred-
itable job of keeping the allies together in the end, albeit at the cost of
what Brent Scowcroft called “a bad strategy” that raised basic ques-
tions about the limits of alliance warfare and about whether the
United States should, in the future, settle instead for coalitions of the
willing, at least in less than the cataclysmic showdowns of the sort
that NATO was initially created to handle.70 One can only wonder
what greater efficiencies might have been registered by a more as-
sertive campaign approach had the U.S. government been willing to
play a more proactive role in leading from the front and setting both
the direction and pace for NATO’s more hesitant allies.71
Lesson One from both Vietnam and Desert Storm should have been
that one must not commit air power in “penny packets,” as the
British say, to play less-than-determined games with the risk calculus
of the other side. Although it can be surgically precise when preci-
sion is called for, air power is, at bottom, a blunt instrument
designed to break things and kill people in pursuit of clear and
militarily achievable objectives. Not without reason have air warfare
professionals repeatedly insisted since Vietnam that if all one wishes
to do is to “send a message,” call Western Union. On this point, Eliot
Cohen summed it up well five years before the Kosovo crisis erupted
when he compared air power’s lately acquired seductiveness to
modern teenage romance in its seeming propensity to offer political
leaders a sense of “gratification without commitment.”72
______________
70John F. Harris, “Berger’s Caution Has Shaped Role of U.S. in War,” Washington Post,
May 16, 1999.
71In a measured indictment of the Clinton administration’s comportment in this re-
gard, two Brookings Institution analysts wrote that “what was missing . . . was less al-
lied will than a demonstrated American ability and willingness to lead a joint effort.
NATO works best when Washington knows what it wants done and leads the effort to
get the alliance there. In the runup to the Kosovo war, both elements were tragically
lacking. . . . Although it is impossible to know whether the allies would have gone
along with a more robust strategy, including early use of ground forces, the United
States never made the case. U.S. policy presumed the allies’ rejection, just as it
presumed congressional opposition to the use of ground forces.” Ivo H. Daalder and
Michael E. O’Hanlon, Winning Ugly: NATO’s War to Save Kosovo, Washington, D.C.,
Brookings Institution, 2000, pp. 98, 222.
72Eliot A. Cohen, “The Mystique of U.S. Air Power,” Foreign Affairs, January/February
1994, p. 109.
To admit that gradualism of the sort applied in Allied Force may be

the wave of the future for any U.S. involvement in coalition warfare
in the years ahead is hardly to accept that it is any more justifiable
from a military point of view for that reason alone. Quite to the con-
trary, the incrementalism of NATO’s air war for Kosovo, right up to
its very end, involved a potential price that went far beyond the loss
of valuable aircraft, munitions, and other expendables for question-
able gain. It risked frittering away the hard-earned reputation for
effectiveness that U.S. air power had finally earned for itself in Desert
Storm after more than three years of unqualified misuse over North
Vietnam a generation earlier. For all his disagreement with so many
other arguments put forward, to no avail, on the proper uses of air
power by his air component commander, General Short, even Gen-
eral Clark emphasized after the air war ended that despite under-
standable pressures for a gradualist approach both from Washington
and among the NATO allies, “once the threshold is crossed to employ
force, then force should be employed as quickly and decisively as
possible. The more rapidly it can be done, the greater the likelihood
of success.” 73
As the Gulf War experience showed, and as both Deliberate Force

and Allied Force ultimately reaffirmed, U.S. air power as it has
evolved since the mid-1970s can do remarkable things when em-
ployed with determination in support of a campaign whose intent is
not in doubt. Yet to conjure up the specter of “air strikes,” NATO or
otherwise, in an effort to project an appearance of “doing some-
thing” without a prior weighing of intended effects or likely conse-
quences is to run the risk of getting bogged down in an operation
with no plausible theory of success. After years of false promises by
its most outspoken prophets, air power has become an unprecedent-
edly capable instrument of force employment in joint warfare. Even
in the best of circumstances, however, it can never be more effective
than the strategy it is intended to support.
______________
73Joseph Fitchett, “Clark Recalls ‘Lessons’ of Kosovo,” International Herald Tribune,
May 3, 2000.
BIBLIOGRAPHY
OFFICIAL PUBLICATIONS
Air Force Basic Doctrine, Maxwell AFB, Alabama, Headquarters Air
Force Doctrine Center, AFDD-1, September 1997.
The Air War Over Serbia: Aerospace Power in Operation Allied Force,
Washington, D.C., Headquarters United States Air Force, April 1,
2000.
Cohen, Secretary of Defense William S., and Chairman of the Joint

Chiefs of Staff General Henry H. Shelton, Kosovo/Operation Allied
Force After-Action Report, Washington, D.C., Department of De-
fense, Report to Congress, January 31, 2000.
Kosovo: Lessons from the Crisis, Report to Parliament by the Secretary

of State for Defense, The Stationery Office, London, England, June
2000.
CONGRESSIONAL TESTIMONY
Clark, General Wesley, USA, testimony to the Senate Armed Services
Committee, Washington, D.C., July 1, 1999.
Cohen, Secretary of Defense William S., and General Henry H. Shel-

ton, “Joint Statement on the Kosovo After-Action Review,” testi-
mony before the Senate Armed Services Committee, Washington,
D.C., October 14, 1999.
251
Esmond, Lieutenant General Marvin R., testimony to the Military

Procurement Subcommittee, House Armed Services Committee,
Washington, D.C., October 19, 1999.
Jumper, General John, USAF, testimony to the Military Readiness

Subcommittee, House Armed Services Committee, Washington,
D.C., October 26, 1999.
BOOKS
Andric, Ivo, The Bridge on the Drina, Chicago, University of Chicago
Press, 1977.
Clark, General Wesley K., Waging Modern War: Bosnia, Kosovo, and
the Future of Combat, New York, Public Affairs, 2001.
Daalder, Ivo H., and Michael E. O’Hanlon, Winning Ugly: NATO’s

War to Save Kosovo, Washington, D.C., Brookings Institution,
2000.
Dylewski, Major General Gary, “The USAF Space Warfare Center:

Bringing Space to the Warfighter,” in Peter L. Hays et al., eds.,
Spacepower for a New Millennium: Space and U.S. National Se-
curity, New York, McGraw-Hill, 2000.
Gentile, Gian P., How Effective Is Strategic Bombing? Lessons Learned

from World War II to Kosovo, New York, New York University
Press, 2001.
Glenny, Misha, The Balkans: Nationalism, War and the Great Powers,
1809–1999, New York: Penguin Books, 2000.
Ignatieff, Michael, Virtual War: Kosovo and Beyond, New York, Henry
Holt and Company, Inc., 2000.
Judah, Tim, Kosovo: War and Revenge, New Haven, Connecticut, Yale
University Press, 2000.
Lambeth, Benjamin S., The Transformation of American Air Power,

Ithaca, New York, Cornell University Press, 2000.
Bibliography 253
Macgregor, Colonel Douglas A., USA, Breaking the Phalanx: A New

Design for Landpower in the 21st Century, Westport, Connecticut,
Praeger, 1997.
Mason, Air Vice Marshal Tony, RAF (Ret.), Air Power: A Centennial
Appraisal, London, Brassey’s, 1994.
Owen, Colonel Robert, USAF, ed., Deliberate Force: A Case Study in

Effective Air Campaigning, Maxwell AFB, Alabama, Air University
Press, January 2000.
Rackham, Peter, ed., Jane’s C4I Systems, 1994–95, London, Jane’s In-
formation Group, 1994.
Short, Lieutenant General Michael C., USAF (Ret.), “An Airman's

Lessons from Kosovo,” in John Andreas Olsen, ed., F r o m
Maneuver Warfare to Kosovo, Trondheim, Norway, Royal
Norwegian Air Force Academy, 2001, pp. 257–288.
Thompson, Wayne, To Hanoi and Back: The U.S. Air Force and North
Vietnam, 1966–1973, Washington, D.C., Smithsonian Institute
Press, 2000.
MONOGRAPHS AND REPORTS

Byman, Daniel L., Matthew C. Waxman, and Eric Larson, Air Power
as a Coercive Instrument, Santa Monica, California, RAND, MR-
1061-AF, 1999.
Hosmer, Stephen T., The Conflict over Kosovo: Why Milosevic

Decided to Settle When He Did, Santa Monica, California, RAND,
MR-1351-AF, 2001.
The Military Balance, 1998/99, London, International Institute for

Strategic Studies, 1998.
Peters, John E., Stuart Johnson, Nora Bensahel, Timothy Liston, and
Traci Williams, European Contributions to Operation Allied Force:
Implications for Transatlantic Cooperation, Santa Monica, Cali-
fornia, RAND, MR-1391-AF, 2001.
Stephens, Alan, Kosovo, or the Future of War, Paper Number 77, Air
Power Studies Center, RAAF Fairbairn, Australia, August 1999.
Strategic Survey 1999/2000, London, England, The International Insti-

tute for Strategic Studies, 2000.
Watts, Barry D., The Military Use of Space: A Diagnostic Assessment,

Washington, D.C., Center for Strategic and Budgetary As-
sessments, February 2001.
White, Lieutenant Colonel Paul K., USAF, Crises After the Storm: An
Appraisal of U.S. Air Operations in Iraq Since the Persian Gulf War,
Military Research Papers No. 2, Washington, D.C., Washington In-
stitute for Near East Policy, 1999.
JOURNAL AND PERIODICAL ARTICLES

Abrams, Elliott, “Just War. Just Means?” National Review, June 28,
1999.
“Air Force Reluctant to Deploy All-Weather Predator UAVs to

Balkans,” Inside the Air Force, April 2, 1999.
Arkin, William M., “Fleet Praises JSOW, Lists Potential Improve-

ments,” Defense Daily, April 26, 2000.
________, “In Praise of Heavy Bombers,” Bulletin of the Atomic

Scientists, July–August 1999.
________, “Kosovo Report Short on Weapons Performance Details,”

Defense Daily, February 10, 2000.
________, “Smart Bombs, Dumb Targeting?” Bulletin of the Atomic

Scientists, May/June 2000.
________, “Top Air Force Leaders to Get Briefed on Serbia Air War
Report,” Defense Daily, June 13, 2000.
Atkinson, David, “B-2s Demonstrated Combat Efficiency over

Kosovo,” Defense Daily, July 1, 1999.
_______, “Stealth Could Play Key Role in Kosovo, Despite Bad

Weather,” Defense Daily, March 23, 1999.
________, and Hunter Keeter, “Apache Role in Kosovo Illustrates

Cracks in Joint Doctrine,” Defense Daily, May 26, 1999.
Bibliography 255
Aubin, Stephen P., “Newsweek and the 14 Tanks,” Air Force Maga-
zine, July 2000.
“B-2 Performed Better in Kosovo Than USAF Expected,” Inside the

Pentagon, July 8, 1999.
Bacevich, Andrew J., “Target Belgrade: Why a Ground War Would Be

a Rout,” National Review, May 3, 1999.
Barry, John, “The Kosovo Cover-Up,” Newsweek, May 15, 2000.
Beaver, Paul, “Mystery Still Shrouds Downing of F-117A Fighter,”

Jane’s Defense Weekly, September 1, 1999.
Ben-Eliahu, Major General Eitan, commander, Israeli Air Force, “Air

Power in the 21st Century: The Impact of Precision Weapons,”
Military Technology, April 2000.
Bender, Roy, “Allies Still Lack Real-Time Retargeting,” Jane’s Defense

Weekly, April 7, 1999.
Betts, Richard K., “Compromised Command: Inside NATO’s First

War,” Foreign Affairs, July/August, 2001.
Bone, Margaret, “Kodak Moments in Kosovo,” The Hook, Spring

2000.
Byman, Daniel L. and Matthew C. Waxman, “Kosovo and the Great

Air Power Debate,” International Security, Spring 2000.
Capaccio, Tony, “JSTARS Led Most Lethal Attacks on Serbs,” Defense

Week, July 6, 1999.
________, “MiGs Tried to Shoot Down Air Force Tanker over Bosnia,”
Defense Week, May 17, 1999.
Cohen, Eliot A., “The Mystique of U.S. Air Power,” Foreign Affairs,
January/February 1994.
Colin, Benoit, and Rene J. Francillon, “L’OTAN en Guerre!” Air Fan,

May 1999.
Corell, John T., “Assumptions Fall in Kosovo,” Air Force Magazine,

June 1999.
________, “Lessons Drawn and Quartered,” Air Force Magazine, De-

cember 1999.
Covault, Craig, “Military Space Dominates Air Strikes,” Aviation

Week and Space Technology, March 29, 1999.
Crawley, Vince, “Air Force Needs 90 Days Between Wars, Chief Says,”
Defense Week, August 9, 1999.
________, “Air Force Secretary Advocates C-130, Predators,” Defense

Week, July 26, 1999.
________, “B-2s See Combat over Yugoslavia,” Defense Week, March

29, 1999.
Cviic, Christopher, “A Victory All the Same,” Survival, Summer 2000.
Daalder, Ivo, “NATO and Kosovo,” The National Interest, Winter

1999/2000.
“Despite Losses, Backers Say Unmanned Systems Excelling Over

Kosovo,” Inside the Pentagon, June 10, 1999.
Donnelly, John, “NRO Chief: Services Ill-Prepared to Work with Spy

Satellites,” Defense Week, July 12, 1999.
Foote, Sheila, “Shelton: Risk Was the Key in Decision Not to Use
Apaches,” Defense Daily, September 10, 1999.
Foreman, Jonathan, “The Casualty Myth,” National Review, May 3,

1999.
Fulghum, David A., “Bomb Shortage Was No Mistake,” Aviation

Week and Space Technology, May 17, 1999.
_______, “Electronic Bombs Darken Belgrade,” Aviation Week and

Space Technology, May 10, 1999.
________, “Joint STARS May Profit from Yugoslav Ops,” Aviation

Week and Space Technology, July 26, 1999.
________, “Kosovo Conflict Spurred New Airborne Technology Use,”

Aviation Week and Space Technology, August 23, 1999.
Bibliography 257
________, “Kosovo Report to Boost New JSF Jamming Role,” Aviation

Week and Space Technology, August 30, 1999.
________, “Lessons Learned May Be Flawed,” Aviation Week and

Space Technology, June 14, 1999.
_________, “NATO Unprepared for Electronic Combat,” Aviation

________, “Report Tallies Damage, Lists U.S. Weaknesses,” Aviation

Week and Space Technology, February 14, 2000.
________, “Russians Analyze U.S. Blackout Bomb,” Aviation Week

and Space Technology, February 14, 2000.
________, “Serb Threat Subsides, but U.S. Still Worries,” Aviation

Week and Space Technology, April 12, 1999.
________, “Yugoslavia Successfully Attacked by Computers,” Aviation

Week and Space Technology, August 23, 1999.
________, and William B. Scott, “Pentagon Gets Lock on F-117 Shoot-

down,” Aviation Week and Space Technology, April 19, 1999.
________, and Robert Wall, “Data Link, EW Problems Pinpointed by

Pentagon,” Aviation Week and Space Technology, September 6,
1999.
________, and Robert Wall, “Intel Mistakes Trigger Chinese Embassy

Bombing,” Aviation Week and Space Technology, May 17, 1999.
Grant, Rebecca, “Air Power Made It Work,” Air Force Magazine,

November 1999.
Grier, Peter, “The Investment in Space,” Air Force Magazine,

February 2000.
Grossman, Elaine M., “Army Commander in Albania Resists Joint

Control over Apache Missions,” Inside the Pentagon, May 20, 1999.
________, “Army’s Cold War Orientation Slowed Apache Deployment

to Balkans,” Inside the Pentagon, May 6, 1999.
________, “As Apaches Near Combat, White House Seeks Diplomatic

Solution,” Inside the Pentagon, May 6, 1999.
________, “Clark’s Firepower Request for Kosovo Prompts Anxiety

Among Chiefs,” Inside the Pentagon, April 15, 1999.
________, “For U.S. Commander in Kosovo, Luck Played Role in

Wartime Success,” Inside the Pentagon, September 9, 1999.
________, “U.S. Military Debates Link Between Kosovo Air War,

Stated Objectives,” Inside the Pentagon, April 20, 2000.
Hagen, William W., “The Balkans’ Lethal Nationalisms,” Foreign Af-

fairs, July/August 1999.
Hammond, Grant T., “Myths of the Air War Over Serbia: Some
‘Lessons’ Not to Learn,” Aerospace Power Journal, Winter 2000.
Harden, Blaine, “The Milosevic Generation,” New York Times Maga-

zine, August 29, 1999.
“‘He Was Calm, Unyielding,’” Newsweek, April 5, 1999.
Hebert, Adam, “Air Force Follows Roadmap in Employment of

Bombers Against Serbia,” Inside the Air Force, April 2, 1999.
Hedges, Chris, “Kosovo’s Next Masters?” Foreign Affairs, May/June

1999.
Hewson, Robert, “Allied Force, Part II: Overwhelming Air Power,”

World Air Power Journal, Winter 1999/2000.
________, “Operation Allied Force: The First 30 Days,” World Air

Power Journal, Fall 1999.
“Hope for the Best, and a Spot of Golf,” The Economist, April 3, 1999.
Hughes, David, “A Pilot’s Best Friend,” Aviation Week and Space

Technology, May 31, 1999.
Ignatieff, Michael, “The Virtual Commander: How NATO Invented a

New Kind of War,” The New Yorker, August 2, 1999.
“Jumper on Air Power,” Air Force Magazine, July 2000.
Kusovac, Zoran, “Russian S-300 SAMs ‘In Serbia,’” Jane’s Defense

Weekly, August 4, 1999.
Bibliography 259
Layne, Christopher, and Benjamin Schwarz, “Kosovo II: For the

Record,” The National Interest, Fall 1999.
Levitin, Oleg, “Inside Moscow’s Kosovo Muddle,” Survival, Spring

2000.
Lorenzo, Ron, “Apache Deployment Has Cost Quarter Billion So

Far,” Defense Week, June 7, 1999.
Luttwak, Edward N., “Give War a Chance,” Foreign Affairs, July/

August 1999.
Mandelbaum, Michael, “A Perfect Failure: NATO’s War Against Yu-

goslavia,” Foreign Affairs, September/October 1999.
McGeary, Johanna, “The Road to Hell,” Time, April 12, 1999.
McPeak, General Merrill A., USAF (Ret.), “The Kosovo Result: The
Facts Speak for Themselves,” Armed Forces Journal International,
September 1999.
Meilinger, Colonel Phillip S., USAF, “Gradual Escalation: NATO’s

Kosovo Air Campaign, Though Decried as a Strategy, May Be the
Future of War,” Armed Forces Journal International, October 1999.
“Missouri-to-Kosovo Flights for B-2 Not a Concern to Wing Com-

mander,” Inside the Air Force, July 2, 1999.
Morrocco, John D., “Kosovo Reveals NATO Interoperability Woes,”

Aviation Week and Space Technology, August 9, 1999.
________, and Robert Wall, “NATO Vows Air Strikes Will Go the
Distance,” Aviation Week and Space Technology, March 29, 1999.
________, David Fulghum, and Robert Wall, “Weather, Weapons

Dearth Slow NATO Strikes,” Aviation Week and Space Technology,
April 5, 1999.
Muradian, Vago, “Stealth Compromised by Not Destroying F-117

Wreckage,” Defense Daily, April 2, 1999.
Muravchik, Joshua, “The Road to Kosovo, Commentary, June 1999.
Murphy, Vice Admiral Daniel J., USN, “The Navy in the Balkans,” Air
Force Magazine, December 1999.
“NATO Jets May Have Erred in Convoy Attack, General Says,”

Aerospace Daily, April 20, 1999.
“NATO’s Game of Chicken,” Newsweek, July 26, 1999.
Nelan, Bruce W., “Into the Fire,” Time, April 5, 1999.
Newman, Richard J., “The Bombs That Failed in Kosovo,” U.S. News
and World Report, September 20, 1999.
________, “In the Skies over Serbia,” U.S. News and World Report,
May 24, 1999.
________, “U.S. Troops Edge Closer to Kosovo,” U.S. News and World
Report, June 7, 1999.
O’Beirne, Kate, “Defenseless: The Military’s Hollow Ring,” National

Review, May 3, 1999.
Pocock, Chris, “Mirage IV Reconnaissance Missions,” World Air

Power Journal, Winter 1999/2000.
Posen, Barry R., “The War for Kosovo: Serbia’s Political-Military

Strategy,” International Security, Spring 2000.
“Ralston Sees Potential for More Wars of Gradual Escalation,” Inside

the Pentagon, September 16, 1999.
“Reporters’ Notebook,” Defense Week, July 19, 1999.
Ripley, Tim, “Harriers over the Kosovo ‘Kill Boxes,’” World Air Power
Journal, Winter 1999/2000.
_______, “‘Serbs Running Out of SAMs,’ Says USA,” Jane’s Defense

Weekly, June 2, 1999.
________, “Tanker Operations,” World Air Power Journal, Winter

1999/2000.
________, “Task Force Hunter,” World Air Power Journal, Winter

1999/2000.
________, “Viper Weasels,” World Air Power Journal, Winter

1999/2000.
Bibliography 261
Roberts, Adam, “NATO’s ‘Humanitarian War’ over Kosovo,” Survival,

Autumn 1999.
Scott, William B., “JTIDS Provides F-15Cs ‘God’s Eye View,’” Aviation
Seigle, Greg, “Prowler Jammers Used to Aid NATO Air Assault,” Jane’s
Defense Weekly, March 31, 1999.
Sharer, Commander Wayne D., USN, “The Navy’s War over Kosovo,”
Proceedings, U.S. Naval Institute, October 1999.
“Shinseki Hints at Restructuring, Aggressive Changes for the Army,”

Inside the Army, June 28, 1999.
Solana, Javier, “NATO’s Success in Kosovo,” Foreign Affairs,

November/December 1999.
Steinberg, James B., “A Perfect Polemic: Blind to Reality on Kosovo,”

Foreign Affairs, November/December 1999.
Stroup, Lieutenant General Theodore G., Jr., USA (Ret.), “Task Force
Hawk: Beyond Expectations,” Army Magazine, August 1999.
“Stumbling into War,” The Economist, March 27, 1999.
Sweetman, Bill, “B-2 Is Maturing into a Fine Spirit,” Jane’s Interna-

tional Defense Review, May 2000.
“Ten Years After,” Proceedings, January 2001.
Thompson, Mark, “Warfighting 101,” Time, June 14, 1999.
Tirpak, John A., “The First Six Weeks,” Air Force Magazine, June 1999.
________, “Kosovo Retrospective,” Air Force Magazine, April 2000.
________, “Lessons Learned and Re-Learned,” Air Force Magazine,

August 1999.
________, “Short’s View of the Air Campaign,” Air Force Magazine,

September 1999.
________, “The State of Precision Engagement,” Air Force Magazine,

March 2000.
Tissue, Lieutenant Colonel Philip C., USMC, “21 Minutes to Bel-

grade,” Proceedings, U.S. Naval Institute, September 1999.
Towell, Pat, “Lawmakers Urge Armed Forces to Focus on High-Tech

Future,” Congressional Quarterly Weekly, June 26, 1999.
“U.S. Mobilizes Guard, Reserve for Balkan Duty,” Air Force Magazine,
June 1999.
“Verbatim Special: The Balkan War,” Air Force Magazine, June 1999.
Wall, Robert, “Airspace Control Challenges Allies,” Aviation Week

and Space Technology, April 26, 1999.
________, “E-2Cs Become Battle Managers with Reduced AEW Role,”

Aviation Week and Space Technology, May 10, 1999.
________, “Joint STARS Changes Operational Scheme,” Aviation

________, “Maverick Fix Tested in Kosovo,” Aviation Week and Space

Technology, September 6, 1999.
________, “NATO Shifts Tactics to Attack Ground Forces,” Aviation

________, “SEAD Concerns Raised in Kosovo,” Aviation Week and

Space Technology, June 26, 1999.
________, “Sustained Carrier Raids Demonstrate New Strike Tactics,”

Aviation Week and Space Technology, May 10, 1999.
“Washington Outlook,” Aviation Week and Space Technology, May 3,

1999.
________, Aviation Week and Space Technology, May 24, 1999.
________, Aviation Week and Space Technology, June 7, 1999.
________, Aviation Week and Space Technology, August 23, 1999.
________, Aviation Week and Space Technology, September 20, 1999.
Wass de Czege, Brigadier General Huba, USA (Ret.), and Lieutenant

Colonel Antulio J. Echevarria II, USA, “Precision Decisions: To
Bibliography 263
Build a Balanced Force, the QDR Might Consider These Four

Propositions,” Armed Forces Journal International, October 2000.
“Wesley Clark Looks Back,” National Journal, February 26, 2000.
NEWSPAPER ARTICLES
Andrews, Edmund L., “Aboard Advanced Radar Flight, U.S. Watches
Combat Zone,” New York Times, June 14, 1999.
Apple, R. W., Jr., “With Decision to Attack, a New Set of U.S. Goals,”
New York Times, March 25, 1999.
Arkin, William M.,“How Sausage Is Made,” Washington Post, July 17,

2000.
________, “Inside the Air Force, Officers Are Frustrated About the Air
War,” Washington Post, April 25, 1999.
________, “Limited Warfare in Kosovo Not Working,” Seattle Times,

May 22, 1999.
Babington, Charles, “Clinton Sticks with Strikes as Poll Shows 51

Percent in U.S. Approve,” Washington Post, March 30, 1999.
________, and Helen Dewar, “President Pleads for Support,”

________, and William Drozdiak, “Belgrade Faces the 11th Hour,

Again,” Washington Post, March 22, 1999.
Balz, Dan, “U.S. Consensus Grows to Send in Ground Troops,”

Barger, Julian, “Bloody Paper Chain May Link Torture to Milosevic,”

The Guardian, June 18, 1999.
Beaumont, Peter, and Patrick Wintour, “Leaks in NATO—and Plan

Bravo Minus,” London Sunday Observer, July 18, 1999.
Becker, Elizabeth, “Needed on Several Fronts, U.S. Jet Force Is

Strained,” New York Times, April 6, 1999.
________, “They’re Unmanned, They Fly Low, and They Get the Pic-
ture,” New York Times, June 3, 1999.
Benedetto, Richard, “Support Not as High as for Other Strikes,” USA

Today, April 2, 1999.
Bennett, Philip, and Steve Coll, “NATO Warplanes Jolt Yugoslav

Power Grid,” Washington Post, May 25, 1999.
Black, Ian, and John Hooper, “Serb Savagery Exposed,” The

Guardian, June 18, 1999.
Block, Robert, “In Belgrade, Hardship Grows Under Sustained Air As-
sault,” Wall Street Journal, May 12, 1999.
Broder, John M., “Clinton Says Milosevic Hurts Claim to Kosovo,”

________, “How to Lay Doubt Aside and Put the Best Face on a Bad
Week in the Balkans,” New York Times, April 1, 1999.
________, “In Grim Week, Pep Talk from the President,” New York
Brown, Justin, “Why U.S. Bombs Failed to Topple Milosevic,” Chris-

tian Science Monitor, March 24, 2000.
Burkins, Glenn, “Serbs Intensify Effort to Down Allied Warplanes,”

Wall Street Journal, May 28, 1999.
Burns, Robert, “Use of Apache Copters Is Not Expected Soon,”

Philadelphia Inquirer, May 19, 1999.
Butcher, Tim, and Patrick Bishop, “NATO Admits Air Campaign

Failed,” London Daily Telegraph, July 22, 1999.
Carroll, James, “The Truth About NATO’s Air War,” Boston Globe,
June 20, 2000.
Clines, Francis X., “NATO Opens Broad Barrage Against Serbs as

Clinton Denounces ‘Brutal Repression,’” New York Times, March
25, 1999.
Cohen, Secretary of Defense William S., interview, “Milosevic Is Far

Weaker Now,” USA Today, May 14, 1999.
Bibliography 265
Cox, James, “Poll: Mission Isn’t Seen as U.S. Victory,” USA Today,
June 15, 1999.
Deans, Bob, “Pentagon Mum About Air Mission,” European Stars and
Stripes, April 27, 1999.
DeParle, Jason, “Allies’ Progress Remains Unclear as Few Details Are

Made Public,” New York Times, April 5, 1999.
Diamond, John, “Yugoslavia, Iraq Talked Air Defense Strategy,”

Philadelphia Inquirer, March 30, 1999.
Dobbs, Michael, “‘Europe’s Last Dictator’ Digs In,” Washington Post,

April 26, 1999.
________, and Karl Vick, “Air Strikes Kill Scores of Refugees,”

Drozdiak, William, “Air War Commander Says Kosovo Victory Near,”

________, “Allies Need Upgrade, General Says,” Washington Post,

June 20, 1999.
________, “Allies Target Computer, Phone Links,” Washington Post,

May 27, 1999.
________, “NATO Leaders Struggle to Find a Winning Strategy,”

________, “Politics Hampered Warfare, Clark Says,” Washington Post,

July 20, 1999.
________, “Yugoslav Troops Devastated by Attack,” Washington Post,

June 9, 1999.
________, and Bradley Graham, “NATO Frustration Grows as Mission

Falls Short,” Washington Post, April 8, 1999.
________, and Steven Mufson, “NATO Sending Tough Terms to Bel-

grade,” Washington Post, June 2, 1999.
________, and Anne Swardson, “Military, Diplomatic Offensives

Bring About Accord,” Washington Post, June 4, 1999.
Eisman, Dale, “Kosovo Lesson: Navy Says It Needs More High-Tech

Tools,” Norfolk Virginian-Pilot, June 10, 1999.
________, “Over Balkans, It’s Beauty vs. the Beast,” Norfolk Virginian-
Pilot, April 26, 1999.
Erlanger, Steven, “Economists Find Bombing Cuts Yugoslavia’s

Production in Half,” New York Times, April 30, 1999.
________, “NATO Was Closer to Ground War in Kosovo Than Is

Widely Realized,” New York Times, November 7, 1999.
________, “U.S. Issues Appeal to Serbs to Halt Attack in Kosovo,” New

Evans, Michael, “Ground War ‘Error,’” London Times, March 24,

2000.
________, “Serb Army Talks of Peace as Armor Takes a Pounding,”

London Times, June 2, 1999.
Fitchett, Joseph, “Clark Recalls ‘Lessons’ of Kosovo,” International

Herald Tribune, May 3, 2000.
________, “For NATO, Keeping Peak Air Traffic on the Go Was a

Critical Goal,” International Herald Tribune, March 31, 2000.
________, “NATO Misjudged Bombing Damage,” International Her-

ald Tribune, June 23, 1999.
Gellman, Barton, “Key Sites Pounded for 2nd Day,” Washington Post,
March 26, 1999.
________, and William Drozdiak, “Conflict Halts Momentum for

Broader Agenda,” Washington Post, June 6, 1999.
“General Admits NATO Exaggerated Bombing Success,” London

Gerstenzang, James, and Elizabeth Shogren, “Serb TV Airs Footage of

3 Captured U.S. Soldiers,” Los Angeles Times, April 1, 1999.
Gertz, Bill, “Remote Radar Allows Serbs to Keep Firing at NATO Jets,”
Washington Times, April 13, 1999.
Bibliography 267
________, and Rowan Scarborough, “Dangerous Drawdown,” Wash-

ington Times, April 30, 1999.
________, and Rowan Scarborough, “Inside the Ring,” Washington

________, and Rowan Scarborough, “No War,” Washington Times,

April 16, 1999.
Gilligan, Andrew, “Russia, Not Bombs, Brought End to War in

Kosovo, Says Jackson,” London Sunday Telegraph, August 1, 1999.
Gordon, Michael R., “Allied Air Chief Stresses Hitting Belgrade Sites,”
New York Times, May 13, 1999.
________, “NATO to Hit Serbs from 2 More Sides,” New York Times,
May 11, 1999.
________, “NATO Says Serbs, Fearing Land War, Dig In on Border,”

________, “A War out of the Night Sky: 10 Hours with a Battle Team,”
________, and Eric Schmitt, “Shift in Targets Let NATO Jets Tip the
Balance,” New York Times, June 5, 1999.
Graham, Bradley, “Air Power ‘Effective, Successful,’ Cohen Says,”

________, “Bombing Spreads,” Washington Post, March 29, 1999.
________, “General Says U.S. Readiness Is Ailing,” Washington Post,

April 30, 1999.
________, “Joint Chiefs Doubted Air Strategy,” Washington Post, April

5, 1999.
________, “Report Says NATO Bombing Killed 500 Civilians in Yu-

goslavia,” Washington Post, February 7, 2000.
________, “U.S. Analysts Misread, Relied on Outdated Maps,” Wash-

________, and William Drozdiak, “Allied Action Fails to Stop Serb

Brutality,” Washington Post, March 31, 1999.
________, and John Lancaster, “Most NATO Bombing Raids Target

Previously Hit Sites,” Washington Post, April 21, 1999.
________, and Dana Priest, “Allies to Begin Flying Refugees Abroad,”

________, and Dana Priest, “‘No Way to Fight a War’: Hard Lessons of
Air Power, Coalitions,” Washington Post, June 6, 1999.
“Ground Troops Lauded,” European Stars and Stripes, August 6,

1999.
Gurdon, Hugo, “France Kept in Dark by Allies,” London Daily Tele-

graph, April 9, 1999.
________, “U.S. Admits Milosevic Spies Are Inside NATO,” London

Daily Telegraph, April 15, 1999.
Harris, John F., “Berger’s Caution Has Shaped Role of U.S. in War,”
________, “Clinton Saw No Alternative to Airstrikes,” Washington

________, “Clinton Says He Might Send Ground Troops,” Washington

Post, May 19, 1999.
________, “Reassuring Rhetoric, Reality in Conflict,” Washington

________, and Bradley Graham, “Clinton Is Reassessing Sufficiency of

Air War,” Washington Post, June 3, 1999.
Havemann, Joel, “Convoy Deaths May Undermine Moral Authority,”

Los Angeles Times, April 15, 1999.
Hedges, Chris, “Angry Serbs Hear a New Explanation: It’s All Russia’s
Fault,” New York Times, July 16, 1999.
Hoffman, Lisa, “U.S. Taxpayers Faced with Mounting Kosovo War

Costs,” Washington Times, June 10, 1999.
Bibliography 269
Kaminski, Matthew, and John Reed, “NATO Link to KLA Rebels May
Have Helped Seal Victory,” Wall Street Journal, July 6, 1999.
Keegan, John, “Modern Weapons Hit War Wisdom,” Sydney Morning

Herald, June 5, 1999.
King, Neil, Jr., “War Against Yugoslavia Lapses into Routine, but
Clock Is Ticking,” Wall Street Journal, May 6, 1999.
Krepinevich, Andrew, “Two Cheers for Air Power,” Wall Street Jour-
nal, June 11, 1999.
Lippman, Thomas W., and Bradley Graham, “NATO Chief Asks

Review of Invasion Planning,” Washington Post, April 22, 1999.
Lippman, Thomas W., and Bradley Graham, “Yugoslavs Fire on U.S.

Troops; 3 Missing,” Washington Post, April 1, 1999.
Lippman, Thomas W., and Dana Priest, “NATO Builds Firepower for
24-Hour Attacks,” Washington Post, March 30, 1999.
Loeb, Vernon, and Steven Mufson, “CIA Analyst Raised Alert on Chi-
na’s Embassy,” Washington Post, June 24, 1999.
Marshall, Tyler, and Richard Boudreaux, “Crisis in Yugoslavia: How

an Uneasy Alliance Prevailed,” Los Angeles Times, June 6, 1999.
McManus, Doyle, “Clinton’s Massive Ground Invasion That Almost

Was,” Los Angeles Times, June 9, 2000.
Miller, Marjorie, “KLA Vows to Disarm If NATO Occupies Kosovo,”

Los Angeles Times, June 7, 1999.
Moniz, Dave, “Eye-to-Eye with a New Kind of War,” Christian Science

Monitor, March 23, 2000.
Moore, Molly, and Bradley Graham, “NATO Plans for Peace, Not
Ground Invasion,” Washington Post, May 17, 1999.
Morin, Richard, “Poll Shows Most Americans Want Negotiations on

Kosovo,” Washington Post, May 18, 1999.
Myers, Steven Lee, “Chinese Embassy Bombing: A Wide Net of

Blame,” New York Times, April 17, 2000.
________, “Damage to Serb Military Less Than Expected,” New York

________, “Early Attacks Focus on Web of Air Defense,” New York

________, “Pentagon Said to Be Adding 300 Planes to Fight Serbs,”

“NATO Chief: Targeting Goals Leaked to Yugoslavia,” Pacific Stars

and Stripes, August 13, 1999.
“A New Bomb Damage Report,” Newsweek, December 20, 1999, p. 4.
Pearlstein, Steven, “NATO Bomb Said to Hit Belgrade Hospital,”

Peltz, James, and Jeff Leeds, “Stealth Fighter’s Crash Reveals a De-
sign’s Limits,” Los Angeles Times, March 30, 1999.
Perlez, Jane, “Clinton and the Joint Chiefs to Discuss Ground

Invasion,” New York Times, June 2, 1999.
________, “For Albright’s Mission, More Problems and Risk,” New

York Times, June 7, 1999.
________, “Holbrooke to Meet Milosevic in Final Peace Effort,” New

________, “Serbs Try to Empty Disputed Province, NATO Aides As-

sert,” New York Times, March 29, 1999.
________, “U.S. Option: Air Attacks May Prove Unpalatable,” New

Pfaff, William, “After NATO’s Lies About Kosovo, It’s Time to Come
Clean,” International Herald Tribune, May 11, 2000.
Priest, Dana, “A Decisive Battle That Never Was,” Washington Post,

September 19, 1999.
________, “France Acted as Group Skeptic,” Washington Post,

September 20, 1999.
Bibliography 271
________, “NATO Pilots Set to Confront Potent Foe,” Washington

________, “NATO Unlikely to Alter Strategy,” Washington Post, March

26, 1999.
________, “Target Selection Was Long Process,” Washington Post,

September 20, 1999.
________, “Tension Grew with Divide in Strategy,” Washington Post,

September 21, 1999.
________, and Peter Finn, “NATO Gives Air Support to Kosovo

Guerrillas,” Washington Post, June 2, 1999.
Raum, Tom, “Cohen: NATO Process Prolonged Air Strikes,” Philadel-

phia Inquirer, July 21, 1999.
Rice, Donald B., “No Stealth to Pentagon’s Bias Against the B-2,” Los
Angeles Times, May 9, 1999.
Richter, Paul, “Air-Only Campaign Offers a False Sense of Security,

Some Say,” Los Angeles Times, June 4, 1999.
________, “B-2 Drops Its Bad PR in Air War,” Los Angeles Times, July 8,
1999.
________, “Bunker-Busters Aim at Heart of Leadership,” Los Angeles

Times, May 5, 1999.
________, “Milosevic War Machine Has a Lot of Fight Left,” Los Ange-
les Times, April 29, 1999.
________, “Officials Say NATO Pounded Milosevic into Submission,”

Los Angeles Times, June 5, 1999.
________, “Time Is Not on the Side of U.S., Allies,” Los Angeles Times,
March 25, 1999.
________, “U.S. Pilots Face Perilous Task, Pentagon Says,” Los Angeles
________, “U.S. Study of War on Yugoslavia Aimed at Boosting Per-

formance,” Los Angeles Times, July 10, 1999.
________, “Use of Ground Troops Not Fully Ruled Out,” Los Angeles
________, and John-Thor Dahlburg, “NATO Broadens Its Battle Strat-

egy,” Los Angeles Times, March 24, 1999.
________, and Lisa Getter, “Mechanical Error, Pilot Error Led to

Apache Crashes,” Los Angeles Times, May 13, 1999.
________, and Doyle McManus, “Pentagon to Tighten Targeting Pro-

cedures,” Los Angeles Times, May 11, 1999.
Ricks, Thomas E., “Why the U.S. Army Is Ill-Equipped to Move

Troops Quickly into Kosovo,” Wall Street Journal, April 16, 1999.
________, and Anne Marie Squeo, “Kosovo Campaign Showcased the

Effectiveness of Air Power,” Wall Street Journal, June 4, 1999.
Robbins, Carla Anne, and Thomas E. Ricks, “Time Is Running Out If

Invasion Is to Remain Option Before Winter,” Wall Street Journal,
May 21, 1999.
Robbins, Carla Anne, Thomas E. Ricks, and Neil King, Jr.,

“Milosevic’s Resolve Spawned Unity, Wider Bombing List in NATO
Alliance,” Wall Street Journal, April 27, 1999.
Ryan, General Michael E., USAF, “Air Power Is Working in Kosovo,”

“Sacked Yugoslav Air Chief Killed,” London Times, June 2, 1999.
Scarborough, Rowan, “Air Force Search and Rescue Operations

Called ‘Broken,’” Washington Times, September 13, 1999.
________, “Apaches Were Sent to Scare Serbs,” Washington Times,

May 21, 1999.
________, “As Strikes Mount, So Do Errors,” Washington Times, May

11, 1999.
________, “Kosovo Target Data Stalled in Transit,” Washington Times,

July 28, 1999.
________, “Military Experts See a Need for Ground Troops,”

Washington Times, March 30, 1999.
Bibliography 273
________, “Momentum for Troops Growing,” Washington Times,

April 5, 1999.
________, “Officers Criticize Air War Strategy,” Washington Times,

May 10, 1999.
________, “Pentagon Intends to Issue Final Count of Serbian Losses,”

Washington Times, July 9, 1999.
________, “Record Deployments Take Toll on Military,” Washington

________, “Smaller U.S. Military Is Spread Thin,” Washington Times,

March 31, 1999.
________, “U.S. Pilots Call NATO Targeting a ‘Disgrace,’” Washington

Schmemann, Serge, “Now, Onward to the Next Kosovo. If There Is

One,” New York Times, June 16, 1999.
Schmitt, Eric, “Aim, Not Arms, at the Root of Mistaken Strike on Em-
bassy,” New York Times, May 10, 1999.
________, “New Army Chief Seeks More Agility and Power,” New York
________, “Pentagon Admits Its Maps of Belgrade Are Out of Date,”

________, “Shrewd Serb Tactics Downed Stealth Jet, U.S. Inquiry

Shows,” New York Times, April 11, 1999.
________, “Weak Serb Defense Puzzles NATO,” New York Times,

March 26, 1999.
________, and Michael R. Gordon, “British Pressing Partners to De-

ploy Ground Troops,” New York Times, May 18, 1999.
________, and Steven Lee Myers, “NATO Said to Focus Raids on Serb
Elite’s Property,” New York Times, April 19, 1999.
Sciolino, Elaine and Ethan Bronner, “How a President, Distracted by

Scandal, Entered Balkan War,” New York Times, April 18, 1999.
Smith, R. Jeffrey, “Belgrade Rebuffs Final U.S. Warning,” Washington

________, and Molly Moore, “Plan for Kosovo Pullout Signed,” Wash-
ington Post, June 10, 1999.
________, and Dana Priest, “Yugoslavia Near Goals in Kosovo,”

Stanley, Alessandra, “Albanian Fighters Say They Aid NATO in Spot-

ting Serb Targets,” New York Times, April 2, 1999.
Suro, Roberto, and Thomas E. Ricks, “Pentagon: Kosovo Air War Data
Leaked,” Washington Post, March 10, 2000.
Taylor, Humphrey, “Win in Kosovo and the Public Will Approve,”

Wall Street Journal, June 3, 1999.
Walker, Tom, “Bomb Video Took Fight out of Milosevic,” London

Sunday Times, January 30, 2000.
Weiner, Tim, “From President, Victory Speech and a Warning,” New

York Times, June 11, 1999.
Whitney, Craig R., “NATO Chief Admits Bombs Fail to Stem Serb
Operations,” New York Times, April 28, 1999.
________, “NATO Had Signs Its Strategy Would Fail Kosovars,” New
________, “On 7th Day, Serb Resilience Gives NATO Leaders Pause,”
Williams, Daniel, “Yugoslavs Yield to NATO Terms,” Washington

Post, June 4, 1999.
________, and Bradley Graham, “Milosevic Admits to Losses of

Personnel,” Washington Post, May 13, 1999.
Wilson, George C., “Memo Says Apaches, Pilots Were Not Ready,”
European Stars and Stripes, June 20, 1999.
Wintour, Patrick, and Peter Beaumont, “Revealed: The Secret Plan to

Invade Kosovo,” London Sunday Observer, July 18, 1999.
Bibliography 275
“Yugoslav Army Lost 524 Soldiers, Top General Says,” International

Herald Tribune, July 22, 1999.
BRIEFINGS
Baldazzi, Colonel E., Italian Air Force, “Host Nation Support for the
Kosovo Air Campaign,” briefing at a conference on “The NATO
Joint Force Air Component Commander Concept in Light of the
Kosovo Air Campaign,” Headquarters NATO Reaction Force Air
Staff, Kalkar, Germany, December 1–3, 1999.
Bivins, Colonel Robert, USAF, Director of Operations, U.S. Air Force

Space Warfare Center, “Space Support to Operation Allied Force:
Preliminary Lessons Learned,” Schriever AFB, Colorado, February
25, 2000.
Clark, General Wesley, SACEUR, Brussels, NATO Headquarters, April

13, 1999.
Corley, Brigadier General John, USAF, news briefing, Office of the

Assistant Secretary of Defense (Public Affairs), the Pentagon,
Washington, D.C., May 8, 2000.
Crawford, Natalie, and others, “USAF EW Management Process

Study,” October 1, 1999.
Dallager, Major General John, USAF, “NATO JFACC Doctrine,”

briefing at a conference on “The NATO Joint Force Air Component
Commander Concept in Light of the Kosovo Air Campaign,”
Headquarters NATO Reaction Force Air Staff, Kalkar, Germany,
December 1–3, 1999.
Ellis, Admiral James O., USN, commander in chief, U.S. Naval Forces,
Europe, and commander, Allied Forces Southern Europe and Joint
Task Force Noble Anvil, “The View from the Top,” 1999.
Hinson, Major General Robert, USAF, commander, 14th Air Force,

“Space Doctrine Lessons from Operation Allied Force,” command
briefing, Vandenberg AFB, California, December 16, 1999.
Leaf, Brigadier General Daniel, USAF, videotaped press statement,

Brussels, Belgium, NATO Office of Information and Press, April 19,
1999.
Pitotti, Colonel Steve, USAF, “Global Environments, Threats, and

Military Strategy (GETM) Update,” Air Armament Summit 2000
briefing, 2000.
Robertson, General Charles T., Jr., USAF, commander in chief, U.S.

Transportation Command, and commander, Air Mobility Com-
mand, “Air War Over Serbia: A Mobility Perspective,” 2000.
Watts, Barry D., “The EA-6B, E-8C, and B-2 in Operation Allied
Force,” Northrop Grumman Analysis Center, Rosslyn, Virginia,
May 8, 2000.

NATO's Air War For Kosovo - A Strategic and Operational Assessment-Rand Publishing (2001)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

NATO's Air War For Kosovo - A Strategic and Operational Assessment-Rand Publishing (2001)

Uploaded by

Copyright:

Available Formats

RAND MR1365-1.

Zubin Potok Medvedja l

Suva Reka Urosevac

On March 24, 1999, NATO embarked on a 78-day air war aimed at

With a view toward capturing the many useful insights to be ex-

This book examines the conduct and results of Operation Allied

Other documents published in this series currently include the fol-

MR-1279-AF, Command and Control and Battle Management: Expe-

MR-1326-AF, Aircraft Weapon Employment in Operation Allied Force,

MR-1351-AF, The Conflict over Kosovo: Why Milosevic Decided to

MR-1391-AF, European Contributions to Operation Allied Force:

Topics examined in series documents nearing completion include:

• Supporting Expeditionary Aerospace Forces: Lessons from the

Project AIR FORCE

2.1. Allied Force Area of Operations .................................... 13

WHY MILOSEVIC GAVE UP WHEN HE DID

NATO ground invasion. Whatever NATO’s declared stance on the

In contrast, Milosevic was living with the daily reality of an increas-

FRICTION AND OPERATIONAL PROBLEMS

problems that were encountered along the way. Among those

• Assessed deficiencies in the suppression of enemy air defenses

In contrast to the far more satisfying SEAD experience in Desert

Still another disappointment centered on what turned out to be

emy troops operating in Kosovo. That disappointment underscored

Pressures to avoid civilian casualties and unintended damage to

dency of allied aircrews to maneuver their aircraft in such a way as to

LAPSES IN STRATEGY AND IMPLEMENTATION

this multistage and circuitous process, two parallel but separate

sisted, as was his command prerogative, upon concentrating on elu-

Despite the success of Allied Force in the end, one misjudgment of

NATO’S AIR WAR IN PERSPECTIVE

After the bombing ended, the predominant tendency among most

The two most important accomplishments of the air war occurred at

Despite these accomplishments, there were enough disappoint-

realm of strategy and execution. First, despite its successful outcome

cal capital needed to develop and enforce a strategy that maximizes

On the plus side, the operation’s successful outcome—despite its

One of the most important realizations to emerge from Allied Force

• A failure to exploit air power’s inherent shock potential and to

These problems identified by the Allied Force experience suggest an

ON THE USES AND ABUSES OF AIR POWER

As the Gulf War experience showed, and as both Deliberate Force

This study has benefited from insights gained from conversations

I am also indebted to Major General Eitan Ben-Eliahu, then–

Most important in this respect, I thank General Gregory Martin,

Beyond this direct research support, I am grateful to Brigadier Gen-

the challenge of 100-percent target identification from medium alti-

Finally, for their helpful comments on all or parts of an earlier draft, I

Steve Hosmer, Dave Kassing, Rob Mullins, Bruce Nardulli, David

AAA Antiaircraft Artillery

BDA Battle Damage Assessment

DSP Defense Support Program

JAC Joint Analysis Center

NPIC National Photographic Interpretation Center

formance that indicated continued deficiencies in need of attention.

less than uniformly satisfactory performance by allied air power.2

Operation Allied Force, largely prompted by humanitarian concerns,

After winning an election in 1989 in which he played heavily on Serb

In February 1998, in an escalating wave of reprisals against the rear-

mediate halt to the escalating strife in Kosovo. The following month,

In the end, Milosevic assented to negotiations and agreed to permit

Before long, however, the OSCE monitors found themselves watch-

On February 6, the Contact Group (made up of representatives from